modern-pdf-lib 0.28.1 → 0.29.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +9 -7
- package/dist/browser.cjs +17 -1
- package/dist/browser.d.cts +3 -3
- package/dist/browser.d.mts +3 -3
- package/dist/browser.mjs +2 -2
- package/dist/{index-_Onej9Wy.d.cts → index--7OpQy6B.d.cts} +518 -11
- package/dist/index--7OpQy6B.d.cts.map +1 -0
- package/dist/{index-5Fv1-Z-p.d.mts → index-Cai7jC6u.d.mts} +518 -11
- package/dist/index-Cai7jC6u.d.mts.map +1 -0
- package/dist/index.cjs +17 -1
- package/dist/index.d.cts +3 -3
- package/dist/index.d.mts +3 -3
- package/dist/index.mjs +2 -2
- package/dist/{src-BOzRFaEE.mjs → src-Ci9ZjqhP.mjs} +2125 -17
- package/dist/{src-vNbsqoIy.cjs → src-DQ63jYIR.cjs} +2217 -13
- package/package.json +3 -2
- package/dist/index-5Fv1-Z-p.d.mts.map +0 -1
- package/dist/index-_Onej9Wy.d.cts.map +0 -1
|
@@ -1,6 +1,9 @@
|
|
|
1
1
|
const require_pdfDocument = require("./pdfDocument-BaQlOkYr.cjs");
|
|
2
2
|
const require_pdfObjects = require("./pdfObjects-BcPlSI0a.cjs");
|
|
3
|
+
const require_streamDecode = require("./streamDecode-kQ-REV7v.cjs");
|
|
3
4
|
const require_pdfForm = require("./pdfForm-BVS_do95.cjs");
|
|
5
|
+
const require_compressionAnalysis = require("./compressionAnalysis-DwxVCe4E.cjs");
|
|
6
|
+
const require_bridge = require("./bridge-ByvzPu5h.cjs");
|
|
4
7
|
const require_loader = require("./loader-DR7H0XOf.cjs");
|
|
5
8
|
let fflate = require("fflate");
|
|
6
9
|
//#region src/core/incrementalWriter.ts
|
|
@@ -17923,7 +17926,7 @@ function skipWS(data, pos) {
|
|
|
17923
17926
|
* Read a decimal integer from `data` starting at `pos`.
|
|
17924
17927
|
* Returns `{ value, nextPos }`.
|
|
17925
17928
|
*/
|
|
17926
|
-
function readInt(data, pos) {
|
|
17929
|
+
function readInt$1(data, pos) {
|
|
17927
17930
|
let value = 0;
|
|
17928
17931
|
let started = false;
|
|
17929
17932
|
while (pos < data.length) {
|
|
@@ -18472,10 +18475,10 @@ var StreamingPdfParser = class StreamingPdfParser {
|
|
|
18472
18475
|
pos = skipWS(data, pos);
|
|
18473
18476
|
while (pos < data.length) {
|
|
18474
18477
|
if (pos + 7 <= data.length && data[pos] === 116 && data[pos + 1] === 114 && data[pos + 2] === 97 && data[pos + 3] === 105 && data[pos + 4] === 108 && data[pos + 5] === 101 && data[pos + 6] === 114) break;
|
|
18475
|
-
const firstObjResult = readInt(data, pos);
|
|
18478
|
+
const firstObjResult = readInt$1(data, pos);
|
|
18476
18479
|
if (firstObjResult.value === -1) break;
|
|
18477
18480
|
pos = skipWS(data, firstObjResult.nextPos);
|
|
18478
|
-
const countResult = readInt(data, pos);
|
|
18481
|
+
const countResult = readInt$1(data, pos);
|
|
18479
18482
|
if (countResult.value === -1) break;
|
|
18480
18483
|
pos = skipWS(data, countResult.nextPos);
|
|
18481
18484
|
const firstObjNum = firstObjResult.value;
|
|
@@ -18554,8 +18557,8 @@ var StreamingPdfParser = class StreamingPdfParser {
|
|
|
18554
18557
|
}
|
|
18555
18558
|
parseXrefStream(data, offset) {
|
|
18556
18559
|
let pos = skipWS(data, offset);
|
|
18557
|
-
pos = skipWS(data, readInt(data, pos).nextPos);
|
|
18558
|
-
pos = skipWS(data, readInt(data, pos).nextPos);
|
|
18560
|
+
pos = skipWS(data, readInt$1(data, pos).nextPos);
|
|
18561
|
+
pos = skipWS(data, readInt$1(data, pos).nextPos);
|
|
18559
18562
|
if (pos + 3 <= data.length && data[pos] === 111 && data[pos + 1] === 98 && data[pos + 2] === 106) pos += 3;
|
|
18560
18563
|
pos = skipWS(data, pos);
|
|
18561
18564
|
const dictResult = parseLightDict(data, pos);
|
|
@@ -18642,10 +18645,10 @@ var StreamingPdfParser = class StreamingPdfParser {
|
|
|
18642
18645
|
*/
|
|
18643
18646
|
parseObjectDict(data, offset) {
|
|
18644
18647
|
let pos = skipWS(data, offset);
|
|
18645
|
-
const objNumResult = readInt(data, pos);
|
|
18648
|
+
const objNumResult = readInt$1(data, pos);
|
|
18646
18649
|
if (objNumResult.value === -1) return void 0;
|
|
18647
18650
|
pos = skipWS(data, objNumResult.nextPos);
|
|
18648
|
-
const genResult = readInt(data, pos);
|
|
18651
|
+
const genResult = readInt$1(data, pos);
|
|
18649
18652
|
if (genResult.value === -1) return void 0;
|
|
18650
18653
|
pos = skipWS(data, genResult.nextPos);
|
|
18651
18654
|
if (pos + 3 <= data.length && data[pos] === 111 && data[pos + 1] === 98 && data[pos + 2] === 106) pos += 3;
|
|
@@ -18741,10 +18744,10 @@ var StreamingPdfParser = class StreamingPdfParser {
|
|
|
18741
18744
|
while (pos < data.length) {
|
|
18742
18745
|
pos = skipWS(data, pos);
|
|
18743
18746
|
if (pos >= data.length || data[pos] === 93) break;
|
|
18744
|
-
const objResult = readInt(data, pos);
|
|
18747
|
+
const objResult = readInt$1(data, pos);
|
|
18745
18748
|
if (objResult.value === -1) break;
|
|
18746
18749
|
pos = skipWS(data, objResult.nextPos);
|
|
18747
|
-
const genResult = readInt(data, pos);
|
|
18750
|
+
const genResult = readInt$1(data, pos);
|
|
18748
18751
|
if (genResult.value === -1) break;
|
|
18749
18752
|
pos = skipWS(data, genResult.nextPos);
|
|
18750
18753
|
if (pos < data.length && data[pos] === 82) {
|
|
@@ -22590,7 +22593,7 @@ function parseFormattedNumber(text) {
|
|
|
22590
22593
|
* @param options Formatting options.
|
|
22591
22594
|
* @returns The formatted string.
|
|
22592
22595
|
*/
|
|
22593
|
-
function formatNumber(value, options = {}) {
|
|
22596
|
+
function formatNumber$1(value, options = {}) {
|
|
22594
22597
|
const { decimals = 2, thousandsSep = ",", decimalSep = ".", negativeStyle = "minus", currency = "", currencyPrepend = true } = options;
|
|
22595
22598
|
const isNeg = value < 0;
|
|
22596
22599
|
const fixed = Math.abs(value).toFixed(decimals);
|
|
@@ -22625,7 +22628,7 @@ function AFNumber_Format(nDec, sepStyle, negStyle, currStyle, strCurrency, bCurr
|
|
|
22625
22628
|
return (value) => {
|
|
22626
22629
|
const num = parseFormattedNumber(value);
|
|
22627
22630
|
if (Number.isNaN(num)) return value;
|
|
22628
|
-
return formatNumber(num, {
|
|
22631
|
+
return formatNumber$1(num, {
|
|
22629
22632
|
decimals: nDec,
|
|
22630
22633
|
thousandsSep: seps.thousands,
|
|
22631
22634
|
decimalSep: seps.decimal,
|
|
@@ -27838,6 +27841,2111 @@ function extractNodeBytesFromParent(fullBuffer, parent, child) {
|
|
|
27838
27841
|
return fullBuffer.subarray(start, start + child.totalLength);
|
|
27839
27842
|
}
|
|
27840
27843
|
//#endregion
|
|
27844
|
+
//#region src/render/matrix.ts
|
|
27845
|
+
/** The identity transform. */
|
|
27846
|
+
function identity() {
|
|
27847
|
+
return [
|
|
27848
|
+
1,
|
|
27849
|
+
0,
|
|
27850
|
+
0,
|
|
27851
|
+
1,
|
|
27852
|
+
0,
|
|
27853
|
+
0
|
|
27854
|
+
];
|
|
27855
|
+
}
|
|
27856
|
+
/**
|
|
27857
|
+
* Concatenate two transforms: the result applies `m` **first**, then `n`
|
|
27858
|
+
* (i.e. `point · m · n`). This matches the PDF `cm` operator, where the new
|
|
27859
|
+
* CTM is `cmMatrix × CTM`.
|
|
27860
|
+
*/
|
|
27861
|
+
function multiply(m, n) {
|
|
27862
|
+
return [
|
|
27863
|
+
m[0] * n[0] + m[1] * n[2],
|
|
27864
|
+
m[0] * n[1] + m[1] * n[3],
|
|
27865
|
+
m[2] * n[0] + m[3] * n[2],
|
|
27866
|
+
m[2] * n[1] + m[3] * n[3],
|
|
27867
|
+
m[4] * n[0] + m[5] * n[2] + n[4],
|
|
27868
|
+
m[4] * n[1] + m[5] * n[3] + n[5]
|
|
27869
|
+
];
|
|
27870
|
+
}
|
|
27871
|
+
/** Apply a transform to a point, returning the transformed `[x, y]`. */
|
|
27872
|
+
function applyToPoint(m, x, y) {
|
|
27873
|
+
return [m[0] * x + m[2] * y + m[4], m[1] * x + m[3] * y + m[5]];
|
|
27874
|
+
}
|
|
27875
|
+
/** A pure translation transform. */
|
|
27876
|
+
function translation(tx, ty) {
|
|
27877
|
+
return [
|
|
27878
|
+
1,
|
|
27879
|
+
0,
|
|
27880
|
+
0,
|
|
27881
|
+
1,
|
|
27882
|
+
tx,
|
|
27883
|
+
ty
|
|
27884
|
+
];
|
|
27885
|
+
}
|
|
27886
|
+
/**
|
|
27887
|
+
* The magnitude by which `m` scales lengths — the geometric mean of the
|
|
27888
|
+
* x- and y-axis scale factors (`sqrt(|det|)`). Used to map a user-space line
|
|
27889
|
+
* width to device space.
|
|
27890
|
+
*/
|
|
27891
|
+
function meanScale(m) {
|
|
27892
|
+
const det = m[0] * m[3] - m[1] * m[2];
|
|
27893
|
+
return Math.sqrt(Math.abs(det));
|
|
27894
|
+
}
|
|
27895
|
+
//#endregion
|
|
27896
|
+
//#region src/render/interpreter.ts
|
|
27897
|
+
/**
|
|
27898
|
+
* @module render/interpreter
|
|
27899
|
+
*
|
|
27900
|
+
* Executes a parsed PDF content stream through a graphics-state machine,
|
|
27901
|
+
* producing a resolution-independent {@link DisplayList} (page space, y-up).
|
|
27902
|
+
* This is the foundation the rasterizer and Canvas adapter render from.
|
|
27903
|
+
*
|
|
27904
|
+
* Supported operators: graphics state (`q`/`Q`/`cm`/`w`/`J`/`j`/`gs`), path
|
|
27905
|
+
* construction (`m`/`l`/`c`/`v`/`y`/`re`/`h`), path painting
|
|
27906
|
+
* (`f`/`F`/`f*`/`S`/`s`/`B`/`B*`/`b`/`b*`/`n`), clipping (`W`/`W*`), color
|
|
27907
|
+
* (`rg`/`RG`/`g`/`G`/`k`/`K`/`cs`/`CS`/`sc`/`scn`/`SC`/`SCN`), text
|
|
27908
|
+
* (`BT`/`ET`/`Tf`/`Td`/`TD`/`Tm`/`T*`/`Tc`/`Tw`/`Tz`/`TL`/`Ts`/`Tr`/`Tj`/`TJ`/`'`/`"`),
|
|
27909
|
+
* and XObjects (`Do`, recursing into form XObjects). Bézier curves are
|
|
27910
|
+
* flattened to polylines.
|
|
27911
|
+
*
|
|
27912
|
+
* @packageDocumentation
|
|
27913
|
+
*/
|
|
27914
|
+
function initialState$1() {
|
|
27915
|
+
return {
|
|
27916
|
+
ctm: identity(),
|
|
27917
|
+
fill: [
|
|
27918
|
+
0,
|
|
27919
|
+
0,
|
|
27920
|
+
0,
|
|
27921
|
+
255
|
|
27922
|
+
],
|
|
27923
|
+
stroke: [
|
|
27924
|
+
0,
|
|
27925
|
+
0,
|
|
27926
|
+
0,
|
|
27927
|
+
255
|
|
27928
|
+
],
|
|
27929
|
+
lineWidth: 1,
|
|
27930
|
+
lineCap: 0,
|
|
27931
|
+
lineJoin: 0,
|
|
27932
|
+
fillAlpha: 1,
|
|
27933
|
+
strokeAlpha: 1,
|
|
27934
|
+
clip: void 0,
|
|
27935
|
+
font: void 0,
|
|
27936
|
+
fontSize: 0,
|
|
27937
|
+
charSpace: 0,
|
|
27938
|
+
wordSpace: 0,
|
|
27939
|
+
hScale: 1,
|
|
27940
|
+
leading: 0,
|
|
27941
|
+
rise: 0,
|
|
27942
|
+
renderMode: 0
|
|
27943
|
+
};
|
|
27944
|
+
}
|
|
27945
|
+
function cloneState(s) {
|
|
27946
|
+
return { ...s };
|
|
27947
|
+
}
|
|
27948
|
+
function num$2(op) {
|
|
27949
|
+
return typeof op === "number" ? op : 0;
|
|
27950
|
+
}
|
|
27951
|
+
function nameOf(op) {
|
|
27952
|
+
if (op instanceof require_pdfObjects.PdfName) return op.value.replace(/^\//, "");
|
|
27953
|
+
if (typeof op === "string") return op.replace(/^\//, "");
|
|
27954
|
+
}
|
|
27955
|
+
function clamp255(v) {
|
|
27956
|
+
return Math.max(0, Math.min(255, Math.round(v * 255)));
|
|
27957
|
+
}
|
|
27958
|
+
function grayRgba(g) {
|
|
27959
|
+
const v = clamp255(g);
|
|
27960
|
+
return [
|
|
27961
|
+
v,
|
|
27962
|
+
v,
|
|
27963
|
+
v,
|
|
27964
|
+
255
|
|
27965
|
+
];
|
|
27966
|
+
}
|
|
27967
|
+
function rgbRgba(r, g, b) {
|
|
27968
|
+
return [
|
|
27969
|
+
clamp255(r),
|
|
27970
|
+
clamp255(g),
|
|
27971
|
+
clamp255(b),
|
|
27972
|
+
255
|
|
27973
|
+
];
|
|
27974
|
+
}
|
|
27975
|
+
function cmykRgba(c, m, y, k) {
|
|
27976
|
+
const [r, g, b] = require_pdfDocument.cmykToRgb(c, m, y, k);
|
|
27977
|
+
return [
|
|
27978
|
+
clamp255(r),
|
|
27979
|
+
clamp255(g),
|
|
27980
|
+
clamp255(b),
|
|
27981
|
+
255
|
|
27982
|
+
];
|
|
27983
|
+
}
|
|
27984
|
+
/** Interpret `sc`/`scn` numeric components by arity → RGBA. */
|
|
27985
|
+
function componentsToRgba(comps) {
|
|
27986
|
+
if (comps.length === 1) return grayRgba(comps[0]);
|
|
27987
|
+
if (comps.length === 3) return rgbRgba(comps[0], comps[1], comps[2]);
|
|
27988
|
+
if (comps.length === 4) return cmykRgba(comps[0], comps[1], comps[2], comps[3]);
|
|
27989
|
+
}
|
|
27990
|
+
const BEZIER_STEPS = 16;
|
|
27991
|
+
function flattenCubic(out, p0, p1, p2, p3) {
|
|
27992
|
+
for (let i = 1; i <= BEZIER_STEPS; i++) {
|
|
27993
|
+
const t = i / BEZIER_STEPS;
|
|
27994
|
+
const mt = 1 - t;
|
|
27995
|
+
const a = mt * mt * mt;
|
|
27996
|
+
const b = 3 * mt * mt * t;
|
|
27997
|
+
const c = 3 * mt * t * t;
|
|
27998
|
+
const d = t * t * t;
|
|
27999
|
+
out.push(a * p0[0] + b * p1[0] + c * p2[0] + d * p3[0], a * p0[1] + b * p1[1] + c * p2[1] + d * p3[1]);
|
|
28000
|
+
}
|
|
28001
|
+
}
|
|
28002
|
+
/**
|
|
28003
|
+
* Execute a parsed content stream into a {@link DisplayList}.
|
|
28004
|
+
*
|
|
28005
|
+
* @param operators - Output of {@link parseContentStream}.
|
|
28006
|
+
* @param options - Page dimensions and (optional) resources/registry.
|
|
28007
|
+
*/
|
|
28008
|
+
function interpretContentStream(operators, options) {
|
|
28009
|
+
const items = [];
|
|
28010
|
+
const origin = options.origin ?? [0, 0];
|
|
28011
|
+
interpretInto(items, operators, initialState$1(), options, 0);
|
|
28012
|
+
return {
|
|
28013
|
+
items,
|
|
28014
|
+
width: options.width,
|
|
28015
|
+
height: options.height,
|
|
28016
|
+
origin
|
|
28017
|
+
};
|
|
28018
|
+
}
|
|
28019
|
+
const MAX_FORM_DEPTH = 12;
|
|
28020
|
+
function interpretInto(items, operators, startState, options, depth) {
|
|
28021
|
+
const stack = [];
|
|
28022
|
+
let gs = startState;
|
|
28023
|
+
let path = [];
|
|
28024
|
+
let cur = null;
|
|
28025
|
+
let curUser = [0, 0];
|
|
28026
|
+
let curPage = [0, 0];
|
|
28027
|
+
let pendingClip = null;
|
|
28028
|
+
let tm = identity();
|
|
28029
|
+
let tlm = identity();
|
|
28030
|
+
const startSub = (x, y) => {
|
|
28031
|
+
if (cur && cur.length >= 4) path.push({
|
|
28032
|
+
points: cur,
|
|
28033
|
+
closed: false
|
|
28034
|
+
});
|
|
28035
|
+
curUser = [x, y];
|
|
28036
|
+
curPage = applyToPoint(gs.ctm, x, y);
|
|
28037
|
+
cur = [curPage[0], curPage[1]];
|
|
28038
|
+
};
|
|
28039
|
+
const lineTo = (x, y) => {
|
|
28040
|
+
if (!cur) startSub(x, y);
|
|
28041
|
+
curUser = [x, y];
|
|
28042
|
+
curPage = applyToPoint(gs.ctm, x, y);
|
|
28043
|
+
cur.push(curPage[0], curPage[1]);
|
|
28044
|
+
};
|
|
28045
|
+
const finishPath = () => {
|
|
28046
|
+
if (cur && cur.length >= 4) path.push({
|
|
28047
|
+
points: cur,
|
|
28048
|
+
closed: false
|
|
28049
|
+
});
|
|
28050
|
+
const result = path;
|
|
28051
|
+
path = [];
|
|
28052
|
+
cur = null;
|
|
28053
|
+
return result;
|
|
28054
|
+
};
|
|
28055
|
+
const applyClipIfPending = (subs) => {
|
|
28056
|
+
if (pendingClip) {
|
|
28057
|
+
gs = {
|
|
28058
|
+
...gs,
|
|
28059
|
+
clip: subs.map((s) => ({
|
|
28060
|
+
points: s.points.slice(),
|
|
28061
|
+
closed: true
|
|
28062
|
+
}))
|
|
28063
|
+
};
|
|
28064
|
+
pendingClip = null;
|
|
28065
|
+
}
|
|
28066
|
+
};
|
|
28067
|
+
const pushOpen = (closed) => {
|
|
28068
|
+
if (cur && cur.length >= 4) path.push({
|
|
28069
|
+
points: cur,
|
|
28070
|
+
closed
|
|
28071
|
+
});
|
|
28072
|
+
cur = null;
|
|
28073
|
+
};
|
|
28074
|
+
for (const { operator, operands } of operators) switch (operator) {
|
|
28075
|
+
case "q":
|
|
28076
|
+
stack.push(cloneState(gs));
|
|
28077
|
+
break;
|
|
28078
|
+
case "Q":
|
|
28079
|
+
if (stack.length) gs = stack.pop();
|
|
28080
|
+
break;
|
|
28081
|
+
case "cm":
|
|
28082
|
+
gs = {
|
|
28083
|
+
...gs,
|
|
28084
|
+
ctm: multiply([
|
|
28085
|
+
num$2(operands[0]),
|
|
28086
|
+
num$2(operands[1]),
|
|
28087
|
+
num$2(operands[2]),
|
|
28088
|
+
num$2(operands[3]),
|
|
28089
|
+
num$2(operands[4]),
|
|
28090
|
+
num$2(operands[5])
|
|
28091
|
+
], gs.ctm)
|
|
28092
|
+
};
|
|
28093
|
+
break;
|
|
28094
|
+
case "w":
|
|
28095
|
+
gs = {
|
|
28096
|
+
...gs,
|
|
28097
|
+
lineWidth: num$2(operands[0])
|
|
28098
|
+
};
|
|
28099
|
+
break;
|
|
28100
|
+
case "J":
|
|
28101
|
+
gs = {
|
|
28102
|
+
...gs,
|
|
28103
|
+
lineCap: num$2(operands[0]) | 0
|
|
28104
|
+
};
|
|
28105
|
+
break;
|
|
28106
|
+
case "j":
|
|
28107
|
+
gs = {
|
|
28108
|
+
...gs,
|
|
28109
|
+
lineJoin: num$2(operands[0]) | 0
|
|
28110
|
+
};
|
|
28111
|
+
break;
|
|
28112
|
+
case "gs":
|
|
28113
|
+
gs = applyExtGState(gs, nameOf(operands[0]), options);
|
|
28114
|
+
break;
|
|
28115
|
+
case "rg":
|
|
28116
|
+
gs = {
|
|
28117
|
+
...gs,
|
|
28118
|
+
fill: rgbRgba(num$2(operands[0]), num$2(operands[1]), num$2(operands[2]))
|
|
28119
|
+
};
|
|
28120
|
+
break;
|
|
28121
|
+
case "RG":
|
|
28122
|
+
gs = {
|
|
28123
|
+
...gs,
|
|
28124
|
+
stroke: rgbRgba(num$2(operands[0]), num$2(operands[1]), num$2(operands[2]))
|
|
28125
|
+
};
|
|
28126
|
+
break;
|
|
28127
|
+
case "g":
|
|
28128
|
+
gs = {
|
|
28129
|
+
...gs,
|
|
28130
|
+
fill: grayRgba(num$2(operands[0]))
|
|
28131
|
+
};
|
|
28132
|
+
break;
|
|
28133
|
+
case "G":
|
|
28134
|
+
gs = {
|
|
28135
|
+
...gs,
|
|
28136
|
+
stroke: grayRgba(num$2(operands[0]))
|
|
28137
|
+
};
|
|
28138
|
+
break;
|
|
28139
|
+
case "k":
|
|
28140
|
+
gs = {
|
|
28141
|
+
...gs,
|
|
28142
|
+
fill: cmykRgba(num$2(operands[0]), num$2(operands[1]), num$2(operands[2]), num$2(operands[3]))
|
|
28143
|
+
};
|
|
28144
|
+
break;
|
|
28145
|
+
case "K":
|
|
28146
|
+
gs = {
|
|
28147
|
+
...gs,
|
|
28148
|
+
stroke: cmykRgba(num$2(operands[0]), num$2(operands[1]), num$2(operands[2]), num$2(operands[3]))
|
|
28149
|
+
};
|
|
28150
|
+
break;
|
|
28151
|
+
case "sc":
|
|
28152
|
+
case "scn": {
|
|
28153
|
+
const rgba = componentsToRgba(operands.filter((o) => typeof o === "number"));
|
|
28154
|
+
if (rgba) gs = {
|
|
28155
|
+
...gs,
|
|
28156
|
+
fill: rgba
|
|
28157
|
+
};
|
|
28158
|
+
break;
|
|
28159
|
+
}
|
|
28160
|
+
case "SC":
|
|
28161
|
+
case "SCN": {
|
|
28162
|
+
const rgba = componentsToRgba(operands.filter((o) => typeof o === "number"));
|
|
28163
|
+
if (rgba) gs = {
|
|
28164
|
+
...gs,
|
|
28165
|
+
stroke: rgba
|
|
28166
|
+
};
|
|
28167
|
+
break;
|
|
28168
|
+
}
|
|
28169
|
+
case "m":
|
|
28170
|
+
startSub(num$2(operands[0]), num$2(operands[1]));
|
|
28171
|
+
break;
|
|
28172
|
+
case "l":
|
|
28173
|
+
lineTo(num$2(operands[0]), num$2(operands[1]));
|
|
28174
|
+
break;
|
|
28175
|
+
case "c": {
|
|
28176
|
+
if (!cur) startSub(curUser[0], curUser[1]);
|
|
28177
|
+
const p1 = applyToPoint(gs.ctm, num$2(operands[0]), num$2(operands[1]));
|
|
28178
|
+
const p2 = applyToPoint(gs.ctm, num$2(operands[2]), num$2(operands[3]));
|
|
28179
|
+
const p3 = applyToPoint(gs.ctm, num$2(operands[4]), num$2(operands[5]));
|
|
28180
|
+
flattenCubic(cur, curPage, p1, p2, p3);
|
|
28181
|
+
curUser = [num$2(operands[4]), num$2(operands[5])];
|
|
28182
|
+
curPage = p3;
|
|
28183
|
+
break;
|
|
28184
|
+
}
|
|
28185
|
+
case "v": {
|
|
28186
|
+
if (!cur) startSub(curUser[0], curUser[1]);
|
|
28187
|
+
const p2 = applyToPoint(gs.ctm, num$2(operands[0]), num$2(operands[1]));
|
|
28188
|
+
const p3 = applyToPoint(gs.ctm, num$2(operands[2]), num$2(operands[3]));
|
|
28189
|
+
flattenCubic(cur, curPage, curPage, p2, p3);
|
|
28190
|
+
curUser = [num$2(operands[2]), num$2(operands[3])];
|
|
28191
|
+
curPage = p3;
|
|
28192
|
+
break;
|
|
28193
|
+
}
|
|
28194
|
+
case "y": {
|
|
28195
|
+
if (!cur) startSub(curUser[0], curUser[1]);
|
|
28196
|
+
const p1 = applyToPoint(gs.ctm, num$2(operands[0]), num$2(operands[1]));
|
|
28197
|
+
const p3 = applyToPoint(gs.ctm, num$2(operands[2]), num$2(operands[3]));
|
|
28198
|
+
flattenCubic(cur, curPage, p1, p3, p3);
|
|
28199
|
+
curUser = [num$2(operands[2]), num$2(operands[3])];
|
|
28200
|
+
curPage = p3;
|
|
28201
|
+
break;
|
|
28202
|
+
}
|
|
28203
|
+
case "re": {
|
|
28204
|
+
const x = num$2(operands[0]);
|
|
28205
|
+
const y = num$2(operands[1]);
|
|
28206
|
+
const w = num$2(operands[2]);
|
|
28207
|
+
const h = num$2(operands[3]);
|
|
28208
|
+
pushOpen(false);
|
|
28209
|
+
const corners = [
|
|
28210
|
+
applyToPoint(gs.ctm, x, y),
|
|
28211
|
+
applyToPoint(gs.ctm, x + w, y),
|
|
28212
|
+
applyToPoint(gs.ctm, x + w, y + h),
|
|
28213
|
+
applyToPoint(gs.ctm, x, y + h)
|
|
28214
|
+
];
|
|
28215
|
+
path.push({
|
|
28216
|
+
points: corners.flat(),
|
|
28217
|
+
closed: true
|
|
28218
|
+
});
|
|
28219
|
+
cur = null;
|
|
28220
|
+
curUser = [x, y];
|
|
28221
|
+
curPage = corners[0];
|
|
28222
|
+
break;
|
|
28223
|
+
}
|
|
28224
|
+
case "h":
|
|
28225
|
+
pushOpen(true);
|
|
28226
|
+
break;
|
|
28227
|
+
case "f":
|
|
28228
|
+
case "F":
|
|
28229
|
+
case "f*": {
|
|
28230
|
+
const subs = finishPath();
|
|
28231
|
+
emitFill(items, subs, operator === "f*" ? "evenodd" : "nonzero", gs);
|
|
28232
|
+
applyClipIfPending(subs);
|
|
28233
|
+
break;
|
|
28234
|
+
}
|
|
28235
|
+
case "S":
|
|
28236
|
+
case "s": {
|
|
28237
|
+
if (operator === "s") pushOpen(true);
|
|
28238
|
+
const subs = finishPath();
|
|
28239
|
+
emitStroke(items, subs, gs);
|
|
28240
|
+
applyClipIfPending(subs);
|
|
28241
|
+
break;
|
|
28242
|
+
}
|
|
28243
|
+
case "B":
|
|
28244
|
+
case "B*":
|
|
28245
|
+
case "b":
|
|
28246
|
+
case "b*": {
|
|
28247
|
+
if (operator === "b" || operator === "b*") pushOpen(true);
|
|
28248
|
+
const subs = finishPath();
|
|
28249
|
+
emitFill(items, subs, operator === "B*" || operator === "b*" ? "evenodd" : "nonzero", gs);
|
|
28250
|
+
emitStroke(items, subs, gs);
|
|
28251
|
+
applyClipIfPending(subs);
|
|
28252
|
+
break;
|
|
28253
|
+
}
|
|
28254
|
+
case "n":
|
|
28255
|
+
applyClipIfPending(finishPath());
|
|
28256
|
+
break;
|
|
28257
|
+
case "W":
|
|
28258
|
+
pendingClip = "nonzero";
|
|
28259
|
+
break;
|
|
28260
|
+
case "W*":
|
|
28261
|
+
pendingClip = "evenodd";
|
|
28262
|
+
break;
|
|
28263
|
+
case "BT":
|
|
28264
|
+
tm = identity();
|
|
28265
|
+
tlm = identity();
|
|
28266
|
+
break;
|
|
28267
|
+
case "ET": break;
|
|
28268
|
+
case "Tf":
|
|
28269
|
+
gs = {
|
|
28270
|
+
...gs,
|
|
28271
|
+
font: nameOf(operands[0]),
|
|
28272
|
+
fontSize: num$2(operands[1])
|
|
28273
|
+
};
|
|
28274
|
+
break;
|
|
28275
|
+
case "Tc":
|
|
28276
|
+
gs = {
|
|
28277
|
+
...gs,
|
|
28278
|
+
charSpace: num$2(operands[0])
|
|
28279
|
+
};
|
|
28280
|
+
break;
|
|
28281
|
+
case "Tw":
|
|
28282
|
+
gs = {
|
|
28283
|
+
...gs,
|
|
28284
|
+
wordSpace: num$2(operands[0])
|
|
28285
|
+
};
|
|
28286
|
+
break;
|
|
28287
|
+
case "Tz":
|
|
28288
|
+
gs = {
|
|
28289
|
+
...gs,
|
|
28290
|
+
hScale: num$2(operands[0]) / 100
|
|
28291
|
+
};
|
|
28292
|
+
break;
|
|
28293
|
+
case "TL":
|
|
28294
|
+
gs = {
|
|
28295
|
+
...gs,
|
|
28296
|
+
leading: num$2(operands[0])
|
|
28297
|
+
};
|
|
28298
|
+
break;
|
|
28299
|
+
case "Ts":
|
|
28300
|
+
gs = {
|
|
28301
|
+
...gs,
|
|
28302
|
+
rise: num$2(operands[0])
|
|
28303
|
+
};
|
|
28304
|
+
break;
|
|
28305
|
+
case "Tr":
|
|
28306
|
+
gs = {
|
|
28307
|
+
...gs,
|
|
28308
|
+
renderMode: num$2(operands[0]) | 0
|
|
28309
|
+
};
|
|
28310
|
+
break;
|
|
28311
|
+
case "Td":
|
|
28312
|
+
tlm = multiply(translation(num$2(operands[0]), num$2(operands[1])), tlm);
|
|
28313
|
+
tm = tlm;
|
|
28314
|
+
break;
|
|
28315
|
+
case "TD":
|
|
28316
|
+
gs = {
|
|
28317
|
+
...gs,
|
|
28318
|
+
leading: -num$2(operands[1])
|
|
28319
|
+
};
|
|
28320
|
+
tlm = multiply(translation(num$2(operands[0]), num$2(operands[1])), tlm);
|
|
28321
|
+
tm = tlm;
|
|
28322
|
+
break;
|
|
28323
|
+
case "Tm":
|
|
28324
|
+
tm = [
|
|
28325
|
+
num$2(operands[0]),
|
|
28326
|
+
num$2(operands[1]),
|
|
28327
|
+
num$2(operands[2]),
|
|
28328
|
+
num$2(operands[3]),
|
|
28329
|
+
num$2(operands[4]),
|
|
28330
|
+
num$2(operands[5])
|
|
28331
|
+
];
|
|
28332
|
+
tlm = tm;
|
|
28333
|
+
break;
|
|
28334
|
+
case "T*":
|
|
28335
|
+
tlm = multiply(translation(0, -gs.leading), tlm);
|
|
28336
|
+
tm = tlm;
|
|
28337
|
+
break;
|
|
28338
|
+
case "Tj":
|
|
28339
|
+
tm = showText(items, textString(operands[0]), tm, gs);
|
|
28340
|
+
break;
|
|
28341
|
+
case "TJ": {
|
|
28342
|
+
const arr = Array.isArray(operands[0]) ? operands[0] : [];
|
|
28343
|
+
let combined = "";
|
|
28344
|
+
for (const el of arr) if (typeof el === "number") tm = multiply(translation(-el / 1e3 * gs.fontSize * gs.hScale, 0), tm);
|
|
28345
|
+
else {
|
|
28346
|
+
combined += textString(el);
|
|
28347
|
+
tm = showText(items, textString(el), tm, gs);
|
|
28348
|
+
}
|
|
28349
|
+
break;
|
|
28350
|
+
}
|
|
28351
|
+
case "'":
|
|
28352
|
+
tlm = multiply(translation(0, -gs.leading), tlm);
|
|
28353
|
+
tm = tlm;
|
|
28354
|
+
tm = showText(items, textString(operands[0]), tm, gs);
|
|
28355
|
+
break;
|
|
28356
|
+
case "\"":
|
|
28357
|
+
gs = {
|
|
28358
|
+
...gs,
|
|
28359
|
+
wordSpace: num$2(operands[0]),
|
|
28360
|
+
charSpace: num$2(operands[1])
|
|
28361
|
+
};
|
|
28362
|
+
tlm = multiply(translation(0, -gs.leading), tlm);
|
|
28363
|
+
tm = tlm;
|
|
28364
|
+
tm = showText(items, textString(operands[2]), tm, gs);
|
|
28365
|
+
break;
|
|
28366
|
+
case "Do":
|
|
28367
|
+
doXObject(items, nameOf(operands[0]), gs, options, depth);
|
|
28368
|
+
break;
|
|
28369
|
+
default: break;
|
|
28370
|
+
}
|
|
28371
|
+
}
|
|
28372
|
+
function freeze(subs) {
|
|
28373
|
+
return subs.map((s) => ({
|
|
28374
|
+
points: s.points,
|
|
28375
|
+
closed: s.closed
|
|
28376
|
+
}));
|
|
28377
|
+
}
|
|
28378
|
+
function emitFill(items, subs, rule, gs) {
|
|
28379
|
+
if (!subs.length) return;
|
|
28380
|
+
items.push({
|
|
28381
|
+
type: "fill",
|
|
28382
|
+
subpaths: freeze(subs),
|
|
28383
|
+
rule,
|
|
28384
|
+
color: gs.fill,
|
|
28385
|
+
alpha: gs.fillAlpha,
|
|
28386
|
+
clip: gs.clip
|
|
28387
|
+
});
|
|
28388
|
+
}
|
|
28389
|
+
function emitStroke(items, subs, gs) {
|
|
28390
|
+
if (!subs.length) return;
|
|
28391
|
+
items.push({
|
|
28392
|
+
type: "stroke",
|
|
28393
|
+
subpaths: freeze(subs),
|
|
28394
|
+
color: gs.stroke,
|
|
28395
|
+
alpha: gs.strokeAlpha,
|
|
28396
|
+
lineWidth: Math.max(gs.lineWidth * meanScale(gs.ctm), .1),
|
|
28397
|
+
lineCap: gs.lineCap,
|
|
28398
|
+
lineJoin: gs.lineJoin,
|
|
28399
|
+
clip: gs.clip
|
|
28400
|
+
});
|
|
28401
|
+
}
|
|
28402
|
+
/** Decode a content-stream string operand to text (best-effort latin-1/WinAnsi). */
|
|
28403
|
+
function textString(op) {
|
|
28404
|
+
return typeof op === "string" ? op : "";
|
|
28405
|
+
}
|
|
28406
|
+
/** Emit a text run and advance the text matrix by an approximate width. */
|
|
28407
|
+
function showText(items, text, tm, gs) {
|
|
28408
|
+
if (text.length) items.push({
|
|
28409
|
+
type: "text",
|
|
28410
|
+
text,
|
|
28411
|
+
font: gs.font,
|
|
28412
|
+
fontSize: gs.fontSize,
|
|
28413
|
+
transform: multiply(multiply([
|
|
28414
|
+
gs.fontSize * gs.hScale,
|
|
28415
|
+
0,
|
|
28416
|
+
0,
|
|
28417
|
+
gs.fontSize,
|
|
28418
|
+
0,
|
|
28419
|
+
gs.rise
|
|
28420
|
+
], tm), gs.ctm),
|
|
28421
|
+
color: gs.fill,
|
|
28422
|
+
alpha: gs.fillAlpha,
|
|
28423
|
+
renderMode: gs.renderMode,
|
|
28424
|
+
clip: gs.clip
|
|
28425
|
+
});
|
|
28426
|
+
let w = 0;
|
|
28427
|
+
for (const ch of text) w += .5 * gs.fontSize + gs.charSpace + (ch === " " ? gs.wordSpace : 0);
|
|
28428
|
+
return multiply(translation(w * gs.hScale, 0), tm);
|
|
28429
|
+
}
|
|
28430
|
+
function resolve$1(obj, options) {
|
|
28431
|
+
if (obj && obj.kind === "ref" && options.registry) return options.registry.resolve(obj);
|
|
28432
|
+
return obj;
|
|
28433
|
+
}
|
|
28434
|
+
function subDict(resources, key, options) {
|
|
28435
|
+
if (!resources) return void 0;
|
|
28436
|
+
const d = resolve$1(resources.get(key), options);
|
|
28437
|
+
return d && d.kind === "dict" ? d : void 0;
|
|
28438
|
+
}
|
|
28439
|
+
function applyExtGState(gs, name, options) {
|
|
28440
|
+
if (!name) return gs;
|
|
28441
|
+
const egs = subDict(options.resources, "/ExtGState", options);
|
|
28442
|
+
const dict = egs ? resolve$1(egs.get("/" + name), options) : void 0;
|
|
28443
|
+
if (!dict || dict.kind !== "dict") return gs;
|
|
28444
|
+
const d = dict;
|
|
28445
|
+
const ca = resolve$1(d.get("/ca"), options);
|
|
28446
|
+
const CA = resolve$1(d.get("/CA"), options);
|
|
28447
|
+
const LW = resolve$1(d.get("/LW"), options);
|
|
28448
|
+
const next = { ...gs };
|
|
28449
|
+
if (ca && ca.kind === "number") next.fillAlpha = ca.value;
|
|
28450
|
+
if (CA && CA.kind === "number") next.strokeAlpha = CA.value;
|
|
28451
|
+
if (LW && LW.kind === "number") next.lineWidth = LW.value;
|
|
28452
|
+
return next;
|
|
28453
|
+
}
|
|
28454
|
+
function doXObject(items, name, gs, options, depth) {
|
|
28455
|
+
if (!name) return;
|
|
28456
|
+
const xobjects = subDict(options.resources, "/XObject", options);
|
|
28457
|
+
const xobj = xobjects ? resolve$1(xobjects.get("/" + name), options) : void 0;
|
|
28458
|
+
if (!xobj || xobj.kind !== "stream") {
|
|
28459
|
+
items.push({
|
|
28460
|
+
type: "image",
|
|
28461
|
+
name,
|
|
28462
|
+
transform: gs.ctm,
|
|
28463
|
+
alpha: gs.fillAlpha,
|
|
28464
|
+
clip: gs.clip
|
|
28465
|
+
});
|
|
28466
|
+
return;
|
|
28467
|
+
}
|
|
28468
|
+
const stream = xobj;
|
|
28469
|
+
const subtype = stream.dict.get("/Subtype");
|
|
28470
|
+
if ((subtype instanceof require_pdfObjects.PdfName ? subtype.value.replace(/^\//, "") : "") === "Form" && depth < MAX_FORM_DEPTH) {
|
|
28471
|
+
const matrixObj = resolve$1(stream.dict.get("/Matrix"), options);
|
|
28472
|
+
let formCtm = gs.ctm;
|
|
28473
|
+
if (matrixObj && matrixObj.kind === "array") {
|
|
28474
|
+
const m = matrixObj.items.map((o) => o && o.kind === "number" ? o.value : 0);
|
|
28475
|
+
if (m.length === 6) formCtm = multiply(m, gs.ctm);
|
|
28476
|
+
}
|
|
28477
|
+
const formResources = subDict(stream.dict, "/Resources", options) ?? options.resources;
|
|
28478
|
+
try {
|
|
28479
|
+
interpretInto(items, require_compressionAnalysis.parseContentStream(require_streamDecode.decodeStreamData(stream)), {
|
|
28480
|
+
...gs,
|
|
28481
|
+
ctm: formCtm
|
|
28482
|
+
}, {
|
|
28483
|
+
...options,
|
|
28484
|
+
resources: formResources
|
|
28485
|
+
}, depth + 1);
|
|
28486
|
+
} catch {}
|
|
28487
|
+
return;
|
|
28488
|
+
}
|
|
28489
|
+
items.push({
|
|
28490
|
+
type: "image",
|
|
28491
|
+
name,
|
|
28492
|
+
transform: gs.ctm,
|
|
28493
|
+
alpha: gs.fillAlpha,
|
|
28494
|
+
clip: gs.clip
|
|
28495
|
+
});
|
|
28496
|
+
}
|
|
28497
|
+
/**
|
|
28498
|
+
* Interpret a {@link PdfPage}'s content into a {@link DisplayList}, using its
|
|
28499
|
+
* crop/media box for dimensions and its resources/registry for `Do`/`gs`/`Tf`
|
|
28500
|
+
* resolution. The entry point for {@link renderPageToImage} et al.
|
|
28501
|
+
*/
|
|
28502
|
+
function interpretPage(page) {
|
|
28503
|
+
const box = page.getCropBox() ?? page.getMediaBox();
|
|
28504
|
+
return interpretContentStream(require_compressionAnalysis.parseContentStream(page.getContentStream()), {
|
|
28505
|
+
width: box.width,
|
|
28506
|
+
height: box.height,
|
|
28507
|
+
origin: [box.x, box.y],
|
|
28508
|
+
resources: page.getOriginalResources(),
|
|
28509
|
+
registry: page.getRegistry()
|
|
28510
|
+
});
|
|
28511
|
+
}
|
|
28512
|
+
//#endregion
|
|
28513
|
+
//#region src/render/rasterizer.ts
|
|
28514
|
+
function resolveScale(opts) {
|
|
28515
|
+
if (typeof opts.scale === "number") return opts.scale;
|
|
28516
|
+
if (typeof opts.dpi === "number") return opts.dpi / 72;
|
|
28517
|
+
return 1;
|
|
28518
|
+
}
|
|
28519
|
+
/** Rasterize a display list to an RGBA8888 buffer. */
|
|
28520
|
+
function rasterize(dl, opts = {}) {
|
|
28521
|
+
const scale = resolveScale(opts);
|
|
28522
|
+
const fullH = Math.max(1, Math.ceil(dl.height * scale));
|
|
28523
|
+
const region = opts.region ?? {
|
|
28524
|
+
x: 0,
|
|
28525
|
+
y: 0,
|
|
28526
|
+
width: Math.max(1, Math.ceil(dl.width * scale)),
|
|
28527
|
+
height: fullH
|
|
28528
|
+
};
|
|
28529
|
+
const W = Math.max(1, region.width);
|
|
28530
|
+
const H = Math.max(1, region.height);
|
|
28531
|
+
const data = new Uint8Array(W * H * 4);
|
|
28532
|
+
const bg = opts.background ?? [
|
|
28533
|
+
255,
|
|
28534
|
+
255,
|
|
28535
|
+
255,
|
|
28536
|
+
255
|
|
28537
|
+
];
|
|
28538
|
+
if (bg !== "transparent") for (let i = 0; i < data.length; i += 4) {
|
|
28539
|
+
data[i] = bg[0];
|
|
28540
|
+
data[i + 1] = bg[1];
|
|
28541
|
+
data[i + 2] = bg[2];
|
|
28542
|
+
data[i + 3] = bg[3];
|
|
28543
|
+
}
|
|
28544
|
+
const [ox, oy] = dl.origin;
|
|
28545
|
+
/** page space → region-local pixel space (with y-flip + tile offset). */
|
|
28546
|
+
const toPx = (x, y) => [(x - ox) * scale - region.x, fullH - (y - oy) * scale - region.y];
|
|
28547
|
+
const renderText = opts.renderText ?? true;
|
|
28548
|
+
for (const item of dl.items) {
|
|
28549
|
+
const clip = item.clip ? toPixelSubpaths(item.clip, toPx) : void 0;
|
|
28550
|
+
if (item.type === "fill") fillSubpaths(data, W, H, toPixelSubpaths(item.subpaths, toPx), item.rule, item.color, item.alpha, clip);
|
|
28551
|
+
else if (item.type === "stroke") strokeSubpaths(data, W, H, item.subpaths, toPx, item.lineWidth * scale, item.color, item.alpha, clip);
|
|
28552
|
+
else if (item.type === "text" && renderText && item.renderMode !== 3 && item.renderMode !== 7) fillTextBoxes(data, W, H, item, toPx, clip);
|
|
28553
|
+
}
|
|
28554
|
+
return {
|
|
28555
|
+
data,
|
|
28556
|
+
width: W,
|
|
28557
|
+
height: H
|
|
28558
|
+
};
|
|
28559
|
+
}
|
|
28560
|
+
/**
|
|
28561
|
+
* Render a {@link PdfPage} to a PNG image.
|
|
28562
|
+
* @returns The PNG bytes plus pixel dimensions.
|
|
28563
|
+
*/
|
|
28564
|
+
async function renderPageToImage(page, opts = {}) {
|
|
28565
|
+
const img = rasterize(interpretPage(page), opts);
|
|
28566
|
+
return {
|
|
28567
|
+
data: encodePngFromPixels(img.data, img.width, img.height, 4),
|
|
28568
|
+
width: img.width,
|
|
28569
|
+
height: img.height
|
|
28570
|
+
};
|
|
28571
|
+
}
|
|
28572
|
+
function toPixelSubpaths(subpaths, toPx) {
|
|
28573
|
+
return subpaths.map((s) => {
|
|
28574
|
+
const out = [];
|
|
28575
|
+
for (let i = 0; i + 1 < s.points.length; i += 2) {
|
|
28576
|
+
const [px, py] = toPx(s.points[i], s.points[i + 1]);
|
|
28577
|
+
out.push(px, py);
|
|
28578
|
+
}
|
|
28579
|
+
return out;
|
|
28580
|
+
});
|
|
28581
|
+
}
|
|
28582
|
+
function buildEdges(subpaths) {
|
|
28583
|
+
const edges = [];
|
|
28584
|
+
for (const pts of subpaths) {
|
|
28585
|
+
const n = pts.length / 2;
|
|
28586
|
+
if (n < 2) continue;
|
|
28587
|
+
for (let i = 0; i < n; i++) {
|
|
28588
|
+
const x0 = pts[i * 2];
|
|
28589
|
+
const y0 = pts[i * 2 + 1];
|
|
28590
|
+
const j = (i + 1) % n;
|
|
28591
|
+
const x1 = pts[j * 2];
|
|
28592
|
+
const y1 = pts[j * 2 + 1];
|
|
28593
|
+
if (y0 === y1) continue;
|
|
28594
|
+
if (y0 < y1) edges.push({
|
|
28595
|
+
ymin: y0,
|
|
28596
|
+
ymax: y1,
|
|
28597
|
+
xAtYmin: x0,
|
|
28598
|
+
dxdy: (x1 - x0) / (y1 - y0),
|
|
28599
|
+
dir: 1
|
|
28600
|
+
});
|
|
28601
|
+
else edges.push({
|
|
28602
|
+
ymin: y1,
|
|
28603
|
+
ymax: y0,
|
|
28604
|
+
xAtYmin: x1,
|
|
28605
|
+
dxdy: (x0 - x1) / (y0 - y1),
|
|
28606
|
+
dir: -1
|
|
28607
|
+
});
|
|
28608
|
+
}
|
|
28609
|
+
}
|
|
28610
|
+
return edges;
|
|
28611
|
+
}
|
|
28612
|
+
const SUBSAMPLES = 4;
|
|
28613
|
+
function fillSubpaths(data, W, H, subpaths, rule, color, alpha, clip) {
|
|
28614
|
+
const edges = buildEdges(subpaths);
|
|
28615
|
+
if (!edges.length) return;
|
|
28616
|
+
const clipEdges = clip ? buildEdges(clip) : void 0;
|
|
28617
|
+
let yTop = H;
|
|
28618
|
+
let yBot = 0;
|
|
28619
|
+
for (const e of edges) {
|
|
28620
|
+
yTop = Math.min(yTop, e.ymin);
|
|
28621
|
+
yBot = Math.max(yBot, e.ymax);
|
|
28622
|
+
}
|
|
28623
|
+
const rowStart = Math.max(0, Math.floor(yTop));
|
|
28624
|
+
const rowEnd = Math.min(H, Math.ceil(yBot));
|
|
28625
|
+
const coverage = new Float32Array(W);
|
|
28626
|
+
const inc = 1 / SUBSAMPLES;
|
|
28627
|
+
for (let y = rowStart; y < rowEnd; y++) {
|
|
28628
|
+
coverage.fill(0);
|
|
28629
|
+
for (let s = 0; s < SUBSAMPLES; s++) accumulateSpans(coverage, W, edges, y + (s + .5) / SUBSAMPLES, rule, inc, clipEdges);
|
|
28630
|
+
const r = color[0];
|
|
28631
|
+
const g = color[1];
|
|
28632
|
+
const b = color[2];
|
|
28633
|
+
const ca = color[3] / 255 * alpha;
|
|
28634
|
+
const base = y * W * 4;
|
|
28635
|
+
for (let x = 0; x < W; x++) {
|
|
28636
|
+
const cov = coverage[x];
|
|
28637
|
+
if (cov <= 0) continue;
|
|
28638
|
+
blend(data, base + x * 4, r, g, b, Math.min(1, cov) * ca);
|
|
28639
|
+
}
|
|
28640
|
+
}
|
|
28641
|
+
}
|
|
28642
|
+
/** Build winding spans at sub-scanline `sy`, adding fractional coverage. */
|
|
28643
|
+
function accumulateSpans(coverage, W, edges, sy, rule, weight, clipEdges) {
|
|
28644
|
+
const xs = crossings(edges, sy);
|
|
28645
|
+
if (xs.length < 2) return;
|
|
28646
|
+
const clipSpans = clipEdges ? rowSpans(crossings(clipEdges, sy), "nonzero") : void 0;
|
|
28647
|
+
const spans = rowSpans(xs, rule);
|
|
28648
|
+
for (const [x0, x1] of spans) if (clipSpans) for (const [c0, c1] of clipSpans) addSpan(coverage, W, Math.max(x0, c0), Math.min(x1, c1), weight);
|
|
28649
|
+
else addSpan(coverage, W, x0, x1, weight);
|
|
28650
|
+
}
|
|
28651
|
+
function crossings(edges, sy) {
|
|
28652
|
+
const xs = [];
|
|
28653
|
+
for (const e of edges) if (sy >= e.ymin && sy < e.ymax) xs.push({
|
|
28654
|
+
x: e.xAtYmin + (sy - e.ymin) * e.dxdy,
|
|
28655
|
+
dir: e.dir
|
|
28656
|
+
});
|
|
28657
|
+
xs.sort((a, b) => a.x - b.x);
|
|
28658
|
+
return xs;
|
|
28659
|
+
}
|
|
28660
|
+
function rowSpans(xs, rule) {
|
|
28661
|
+
const spans = [];
|
|
28662
|
+
let winding = 0;
|
|
28663
|
+
for (let i = 0; i + 1 < xs.length; i++) {
|
|
28664
|
+
winding += xs[i].dir;
|
|
28665
|
+
if (rule === "evenodd" ? (i + 1) % 2 === 1 : winding !== 0) spans.push([xs[i].x, xs[i + 1].x]);
|
|
28666
|
+
}
|
|
28667
|
+
return spans;
|
|
28668
|
+
}
|
|
28669
|
+
/** Add `weight` coverage over `[x0, x1)` with fractional pixel ends. */
|
|
28670
|
+
function addSpan(coverage, W, x0, x1, weight) {
|
|
28671
|
+
if (x1 <= x0) return;
|
|
28672
|
+
const a = Math.max(0, x0);
|
|
28673
|
+
const b = Math.min(W, x1);
|
|
28674
|
+
if (b <= a) return;
|
|
28675
|
+
const ia = Math.floor(a);
|
|
28676
|
+
const ib = Math.floor(b - 1e-9);
|
|
28677
|
+
if (ia === ib) {
|
|
28678
|
+
coverage[ia] += (b - a) * weight;
|
|
28679
|
+
return;
|
|
28680
|
+
}
|
|
28681
|
+
coverage[ia] += (ia + 1 - a) * weight;
|
|
28682
|
+
for (let x = ia + 1; x < ib; x++) coverage[x] += weight;
|
|
28683
|
+
coverage[ib] += (b - ib) * weight;
|
|
28684
|
+
}
|
|
28685
|
+
function blend(data, idx, r, g, b, a) {
|
|
28686
|
+
if (a <= 0) return;
|
|
28687
|
+
const ia = 1 - a;
|
|
28688
|
+
data[idx] = Math.round(r * a + data[idx] * ia);
|
|
28689
|
+
data[idx + 1] = Math.round(g * a + data[idx + 1] * ia);
|
|
28690
|
+
data[idx + 2] = Math.round(b * a + data[idx + 2] * ia);
|
|
28691
|
+
data[idx + 3] = Math.round(255 * a + data[idx + 3] * ia);
|
|
28692
|
+
}
|
|
28693
|
+
function strokeSubpaths(data, W, H, subpaths, toPx, widthPx, color, alpha, clip) {
|
|
28694
|
+
const hw = Math.max(widthPx / 2, .35);
|
|
28695
|
+
for (const sp of subpaths) {
|
|
28696
|
+
const pts = sp.points;
|
|
28697
|
+
const n = pts.length / 2;
|
|
28698
|
+
const segCount = sp.closed ? n : n - 1;
|
|
28699
|
+
for (let i = 0; i < segCount; i++) {
|
|
28700
|
+
const a = toPx(pts[i * 2], pts[i * 2 + 1]);
|
|
28701
|
+
const j = (i + 1) % n;
|
|
28702
|
+
const b = toPx(pts[j * 2], pts[j * 2 + 1]);
|
|
28703
|
+
const dx = b[0] - a[0];
|
|
28704
|
+
const dy = b[1] - a[1];
|
|
28705
|
+
const len = Math.hypot(dx, dy);
|
|
28706
|
+
if (len < 1e-6) continue;
|
|
28707
|
+
const nx = -dy / len * hw;
|
|
28708
|
+
const ny = dx / len * hw;
|
|
28709
|
+
fillSubpaths(data, W, H, [[
|
|
28710
|
+
a[0] + nx,
|
|
28711
|
+
a[1] + ny,
|
|
28712
|
+
b[0] + nx,
|
|
28713
|
+
b[1] + ny,
|
|
28714
|
+
b[0] - nx,
|
|
28715
|
+
b[1] - ny,
|
|
28716
|
+
a[0] - nx,
|
|
28717
|
+
a[1] - ny
|
|
28718
|
+
]], "nonzero", color, alpha, clip);
|
|
28719
|
+
}
|
|
28720
|
+
}
|
|
28721
|
+
}
|
|
28722
|
+
function fillTextBoxes(data, W, H, item, toPx, clip) {
|
|
28723
|
+
const m = item.transform;
|
|
28724
|
+
const boxes = [];
|
|
28725
|
+
let i = 0;
|
|
28726
|
+
for (const ch of item.text) {
|
|
28727
|
+
if (ch !== " ") {
|
|
28728
|
+
const x0 = i * .5 + .05;
|
|
28729
|
+
const x1 = i * .5 + .45;
|
|
28730
|
+
const src = [
|
|
28731
|
+
[x0, 0],
|
|
28732
|
+
[x1, 0],
|
|
28733
|
+
[x1, .62],
|
|
28734
|
+
[x0, .62]
|
|
28735
|
+
];
|
|
28736
|
+
const flat = [];
|
|
28737
|
+
for (const [tx, ty] of src) {
|
|
28738
|
+
const [ux, uy] = applyToPoint(m, tx, ty);
|
|
28739
|
+
const [px, py] = toPx(ux, uy);
|
|
28740
|
+
flat.push(px, py);
|
|
28741
|
+
}
|
|
28742
|
+
boxes.push(flat);
|
|
28743
|
+
}
|
|
28744
|
+
i++;
|
|
28745
|
+
}
|
|
28746
|
+
if (boxes.length) fillSubpaths(data, W, H, boxes, "nonzero", item.color, item.alpha * .85, clip);
|
|
28747
|
+
}
|
|
28748
|
+
//#endregion
|
|
28749
|
+
//#region src/render/canvas.ts
|
|
28750
|
+
function effectiveScale(opts) {
|
|
28751
|
+
return (typeof opts.scale === "number" ? opts.scale : typeof opts.dpi === "number" ? opts.dpi / 72 : 1) * (opts.pixelRatio ?? 1);
|
|
28752
|
+
}
|
|
28753
|
+
function rgbaStr(c, alpha) {
|
|
28754
|
+
const a = c[3] / 255 * alpha;
|
|
28755
|
+
return `rgba(${c[0]}, ${c[1]}, ${c[2]}, ${Number(a.toFixed(3))})`;
|
|
28756
|
+
}
|
|
28757
|
+
/** Replay a display list onto a 2D context. */
|
|
28758
|
+
function renderDisplayListToCanvas(dl, ctx, opts = {}) {
|
|
28759
|
+
const scale = effectiveScale(opts);
|
|
28760
|
+
const H = dl.height * scale;
|
|
28761
|
+
const [ox, oy] = dl.origin;
|
|
28762
|
+
const toPx = (x, y) => [(x - ox) * scale, H - (y - oy) * scale];
|
|
28763
|
+
const family = opts.fontFamily ?? "sans-serif";
|
|
28764
|
+
for (const item of dl.items) if (item.type === "fill") {
|
|
28765
|
+
ctx.fillStyle = rgbaStr(item.color, item.alpha);
|
|
28766
|
+
ctx.beginPath();
|
|
28767
|
+
tracePath(ctx, item.subpaths, toPx);
|
|
28768
|
+
ctx.fill(item.rule);
|
|
28769
|
+
} else if (item.type === "stroke") {
|
|
28770
|
+
ctx.strokeStyle = rgbaStr(item.color, item.alpha);
|
|
28771
|
+
ctx.lineWidth = Math.max(item.lineWidth * scale, .5);
|
|
28772
|
+
ctx.beginPath();
|
|
28773
|
+
tracePath(ctx, item.subpaths, toPx);
|
|
28774
|
+
ctx.stroke();
|
|
28775
|
+
} else if (item.type === "text" && item.renderMode !== 3 && item.renderMode !== 7) {
|
|
28776
|
+
const m = item.transform;
|
|
28777
|
+
const [px, py] = toPx(...applyToPoint(m, 0, 0));
|
|
28778
|
+
const sizePx = Math.hypot(m[2], m[3]) * scale;
|
|
28779
|
+
if (sizePx >= .5 && item.text.length) {
|
|
28780
|
+
ctx.fillStyle = rgbaStr(item.color, item.alpha);
|
|
28781
|
+
ctx.font = `${Number(sizePx.toFixed(2))}px ${family}`;
|
|
28782
|
+
ctx.fillText(item.text, px, py);
|
|
28783
|
+
}
|
|
28784
|
+
}
|
|
28785
|
+
}
|
|
28786
|
+
function tracePath(ctx, subpaths, toPx) {
|
|
28787
|
+
for (const sp of subpaths) {
|
|
28788
|
+
const pts = sp.points;
|
|
28789
|
+
if (pts.length < 4) continue;
|
|
28790
|
+
const [x0, y0] = toPx(pts[0], pts[1]);
|
|
28791
|
+
ctx.moveTo(x0, y0);
|
|
28792
|
+
for (let i = 2; i + 1 < pts.length; i += 2) {
|
|
28793
|
+
const [x, y] = toPx(pts[i], pts[i + 1]);
|
|
28794
|
+
ctx.lineTo(x, y);
|
|
28795
|
+
}
|
|
28796
|
+
if (sp.closed) ctx.closePath();
|
|
28797
|
+
}
|
|
28798
|
+
}
|
|
28799
|
+
/**
|
|
28800
|
+
* Render a {@link PdfPage} onto a 2D context. The context's canvas should be
|
|
28801
|
+
* sized to `page.width × scale` by `page.height × scale`.
|
|
28802
|
+
*/
|
|
28803
|
+
function renderPageToCanvas(page, ctx, opts = {}) {
|
|
28804
|
+
renderDisplayListToCanvas(interpretPage(page), ctx, opts);
|
|
28805
|
+
}
|
|
28806
|
+
//#endregion
|
|
28807
|
+
//#region src/render/thumbnail.ts
|
|
28808
|
+
/**
|
|
28809
|
+
* @module render/thumbnail
|
|
28810
|
+
*
|
|
28811
|
+
* Generate a small preview PNG of a page by rasterizing it at a scale that fits
|
|
28812
|
+
* the longest side to a target size. (Persisting a `/Thumb` image XObject on the
|
|
28813
|
+
* page — ISO 32000-2 §12.3.4 — requires page-dictionary write access not yet on
|
|
28814
|
+
* the public {@link PdfPage} API; tracked for a later patch.)
|
|
28815
|
+
*
|
|
28816
|
+
* @packageDocumentation
|
|
28817
|
+
*/
|
|
28818
|
+
/**
|
|
28819
|
+
* Render `page` to a thumbnail PNG whose longest side is `maxSize` pixels,
|
|
28820
|
+
* preserving aspect ratio.
|
|
28821
|
+
*/
|
|
28822
|
+
async function generateThumbnail(page, opts = {}) {
|
|
28823
|
+
const maxSize = opts.maxSize ?? 256;
|
|
28824
|
+
const dl = interpretPage(page);
|
|
28825
|
+
const img = rasterize(dl, {
|
|
28826
|
+
scale: maxSize / (Math.max(dl.width, dl.height) || 1),
|
|
28827
|
+
background: opts.background,
|
|
28828
|
+
renderText: opts.renderText ?? true
|
|
28829
|
+
});
|
|
28830
|
+
return {
|
|
28831
|
+
data: encodePngFromPixels(img.data, img.width, img.height, 4),
|
|
28832
|
+
width: img.width,
|
|
28833
|
+
height: img.height
|
|
28834
|
+
};
|
|
28835
|
+
}
|
|
28836
|
+
//#endregion
|
|
28837
|
+
//#region src/render/imageExtract.ts
|
|
28838
|
+
/**
|
|
28839
|
+
* Extract and decode all image XObjects referenced by a page.
|
|
28840
|
+
*
|
|
28841
|
+
* @param page - The page whose `/Resources /XObject` images are extracted.
|
|
28842
|
+
* @returns One {@link ExtractedImage} per successfully decoded image XObject.
|
|
28843
|
+
* Images that cannot be decoded (unsupported filters/colour spaces,
|
|
28844
|
+
* missing WASM, malformed data) are skipped silently.
|
|
28845
|
+
*
|
|
28846
|
+
* @example
|
|
28847
|
+
* ```ts
|
|
28848
|
+
* import { loadPdf } from 'modern-pdf-lib';
|
|
28849
|
+
* import { extractImages } from 'modern-pdf-lib/render';
|
|
28850
|
+
*
|
|
28851
|
+
* const doc = await loadPdf(bytes);
|
|
28852
|
+
* for (const img of extractImages(doc.getPage(0))) {
|
|
28853
|
+
* console.log(img.name, img.width, img.height, img.channels, img.hasAlpha);
|
|
28854
|
+
* }
|
|
28855
|
+
* ```
|
|
28856
|
+
*/
|
|
28857
|
+
function extractImages(page) {
|
|
28858
|
+
const out = [];
|
|
28859
|
+
const resources = page.getOriginalResources();
|
|
28860
|
+
if (!resources) return out;
|
|
28861
|
+
const registry = page.getRegistry();
|
|
28862
|
+
const xObjDict = resolveDict$1(resources.get("/XObject"), registry);
|
|
28863
|
+
if (!xObjDict) return out;
|
|
28864
|
+
for (const [name, value] of xObjDict) {
|
|
28865
|
+
const stream = resolveStream$1(value, registry);
|
|
28866
|
+
if (!stream) continue;
|
|
28867
|
+
const subtype = stream.dict.get("/Subtype");
|
|
28868
|
+
if (!(subtype instanceof require_pdfObjects.PdfName) || subtype.value !== "/Image") continue;
|
|
28869
|
+
try {
|
|
28870
|
+
const extracted = decodeImage(name, stream, registry);
|
|
28871
|
+
if (extracted) out.push(extracted);
|
|
28872
|
+
} catch {
|
|
28873
|
+
continue;
|
|
28874
|
+
}
|
|
28875
|
+
}
|
|
28876
|
+
return out;
|
|
28877
|
+
}
|
|
28878
|
+
/**
|
|
28879
|
+
* Decode a single image XObject stream to interleaved 8-bit pixels and
|
|
28880
|
+
* composite any soft mask. Returns `undefined` when the image cannot be
|
|
28881
|
+
* decoded (e.g. JPEG WASM unavailable, unsupported colour space).
|
|
28882
|
+
* @internal
|
|
28883
|
+
*/
|
|
28884
|
+
function decodeImage(name, stream, registry) {
|
|
28885
|
+
const width = readInt(stream.dict.get("/Width"), 0);
|
|
28886
|
+
const height = readInt(stream.dict.get("/Height"), 0);
|
|
28887
|
+
if (width <= 0 || height <= 0) return void 0;
|
|
28888
|
+
const bitsPerComponent = readInt(stream.dict.get("/BitsPerComponent"), 8);
|
|
28889
|
+
const cs = resolveColorSpace(stream.dict.get("/ColorSpace"), registry);
|
|
28890
|
+
const { filters } = require_streamDecode.getStreamFilters(stream.dict);
|
|
28891
|
+
const isJpeg = filters.includes("DCTDecode");
|
|
28892
|
+
let base;
|
|
28893
|
+
if (isJpeg) base = decodeJpegBase(stream, width, height);
|
|
28894
|
+
else base = interpretSamples(require_streamDecode.decodeStreamData(stream), width, height, bitsPerComponent, cs);
|
|
28895
|
+
if (!base) return void 0;
|
|
28896
|
+
const alpha = extractSoftMask(stream, registry, width, height);
|
|
28897
|
+
if (alpha) {
|
|
28898
|
+
const pixels = compositeAlpha(base.pixels, base.channels, width, height, alpha);
|
|
28899
|
+
return {
|
|
28900
|
+
name: stripSlash(name),
|
|
28901
|
+
width,
|
|
28902
|
+
height,
|
|
28903
|
+
channels: 4,
|
|
28904
|
+
pixels,
|
|
28905
|
+
colorSpace: cs.name,
|
|
28906
|
+
bitsPerComponent,
|
|
28907
|
+
hasAlpha: true
|
|
28908
|
+
};
|
|
28909
|
+
}
|
|
28910
|
+
return {
|
|
28911
|
+
name: stripSlash(name),
|
|
28912
|
+
width,
|
|
28913
|
+
height,
|
|
28914
|
+
channels: base.channels,
|
|
28915
|
+
pixels: base.pixels,
|
|
28916
|
+
colorSpace: cs.name,
|
|
28917
|
+
bitsPerComponent,
|
|
28918
|
+
hasAlpha: false
|
|
28919
|
+
};
|
|
28920
|
+
}
|
|
28921
|
+
/**
|
|
28922
|
+
* Decode a DCTDecode (JPEG) image XObject to interleaved 8-bit pixels.
|
|
28923
|
+
*
|
|
28924
|
+
* `decodeStreamData` passes DCTDecode through unchanged (decoding any outer
|
|
28925
|
+
* filters such as an ASCII85 wrapper first), so the result is the raw JPEG
|
|
28926
|
+
* byte stream handed to the WASM decoder.
|
|
28927
|
+
* @internal
|
|
28928
|
+
*/
|
|
28929
|
+
function decodeJpegBase(stream, width, height) {
|
|
28930
|
+
const decoded = require_bridge.decodeJpegWasm(require_streamDecode.decodeStreamData(stream));
|
|
28931
|
+
if (!decoded) return void 0;
|
|
28932
|
+
const w = decoded.width || width;
|
|
28933
|
+
const h = decoded.height || height;
|
|
28934
|
+
const channels = decoded.channels;
|
|
28935
|
+
if (w <= 0 || h <= 0 || channels <= 0) return void 0;
|
|
28936
|
+
if (decoded.pixels.length < w * h * channels) return void 0;
|
|
28937
|
+
return {
|
|
28938
|
+
pixels: decoded.pixels,
|
|
28939
|
+
channels
|
|
28940
|
+
};
|
|
28941
|
+
}
|
|
28942
|
+
/**
|
|
28943
|
+
* Interpret decoded sample bytes as interleaved 8-bit RGB / grayscale
|
|
28944
|
+
* pixels according to the colour space and bits-per-component.
|
|
28945
|
+
* @internal
|
|
28946
|
+
*/
|
|
28947
|
+
function interpretSamples(data, width, height, bpc, cs) {
|
|
28948
|
+
const pixelCount = width * height;
|
|
28949
|
+
if (cs.name === "Indexed" && cs.palette) return expandIndexed(data, width, height, bpc, cs.palette);
|
|
28950
|
+
const components = cs.components;
|
|
28951
|
+
const samples = readComponentSamples(data, pixelCount * components, bpc);
|
|
28952
|
+
if (!samples) return void 0;
|
|
28953
|
+
if (cs.name === "DeviceGray" || cs.name === "CalGray") {
|
|
28954
|
+
const pixels = new Uint8Array(pixelCount);
|
|
28955
|
+
for (let i = 0; i < pixelCount; i++) pixels[i] = samples[i] ?? 0;
|
|
28956
|
+
return {
|
|
28957
|
+
pixels,
|
|
28958
|
+
channels: 1
|
|
28959
|
+
};
|
|
28960
|
+
}
|
|
28961
|
+
if (cs.name === "DeviceCMYK") {
|
|
28962
|
+
const pixels = new Uint8Array(pixelCount * 3);
|
|
28963
|
+
for (let p = 0; p < pixelCount; p++) {
|
|
28964
|
+
const si = p * 4;
|
|
28965
|
+
const [r, g, b] = require_pdfDocument.cmykToRgb((samples[si] ?? 0) / 255, (samples[si + 1] ?? 0) / 255, (samples[si + 2] ?? 0) / 255, (samples[si + 3] ?? 0) / 255);
|
|
28966
|
+
const di = p * 3;
|
|
28967
|
+
pixels[di] = Math.round(r * 255);
|
|
28968
|
+
pixels[di + 1] = Math.round(g * 255);
|
|
28969
|
+
pixels[di + 2] = Math.round(b * 255);
|
|
28970
|
+
}
|
|
28971
|
+
return {
|
|
28972
|
+
pixels,
|
|
28973
|
+
channels: 3
|
|
28974
|
+
};
|
|
28975
|
+
}
|
|
28976
|
+
const pixels = new Uint8Array(pixelCount * 3);
|
|
28977
|
+
for (let i = 0; i < pixelCount * 3; i++) pixels[i] = samples[i] ?? 0;
|
|
28978
|
+
return {
|
|
28979
|
+
pixels,
|
|
28980
|
+
channels: 3
|
|
28981
|
+
};
|
|
28982
|
+
}
|
|
28983
|
+
/**
|
|
28984
|
+
* Expand an indexed (palette) image to interleaved RGB pixels.
|
|
28985
|
+
* @internal
|
|
28986
|
+
*/
|
|
28987
|
+
function expandIndexed(data, width, height, bpc, palette) {
|
|
28988
|
+
const pixelCount = width * height;
|
|
28989
|
+
const pixels = new Uint8Array(pixelCount * 3);
|
|
28990
|
+
const maxIndex = Math.floor(palette.length / 3) - 1;
|
|
28991
|
+
const indices = readPackedRows(data, width, height, 1, bpc);
|
|
28992
|
+
for (let p = 0; p < pixelCount; p++) {
|
|
28993
|
+
let idx = indices[p] ?? 0;
|
|
28994
|
+
if (idx > maxIndex) idx = maxIndex < 0 ? 0 : maxIndex;
|
|
28995
|
+
const pi = idx * 3;
|
|
28996
|
+
const di = p * 3;
|
|
28997
|
+
pixels[di] = palette[pi] ?? 0;
|
|
28998
|
+
pixels[di + 1] = palette[pi + 1] ?? 0;
|
|
28999
|
+
pixels[di + 2] = palette[pi + 2] ?? 0;
|
|
29000
|
+
}
|
|
29001
|
+
return {
|
|
29002
|
+
pixels,
|
|
29003
|
+
channels: 3
|
|
29004
|
+
};
|
|
29005
|
+
}
|
|
29006
|
+
/**
|
|
29007
|
+
* Read `count` component samples, normalizing sub-byte / 16-bit components
|
|
29008
|
+
* to 8-bit values. Returns `undefined` when the buffer is too small.
|
|
29009
|
+
* @internal
|
|
29010
|
+
*/
|
|
29011
|
+
function readComponentSamples(data, count, bpc) {
|
|
29012
|
+
if (bpc === 8) {
|
|
29013
|
+
if (data.length < count) return void 0;
|
|
29014
|
+
return data.subarray(0, count);
|
|
29015
|
+
}
|
|
29016
|
+
if (bpc === 16) {
|
|
29017
|
+
if (data.length < count * 2) return void 0;
|
|
29018
|
+
const out = new Uint8Array(count);
|
|
29019
|
+
for (let i = 0; i < count; i++) out[i] = data[i * 2] ?? 0;
|
|
29020
|
+
return out;
|
|
29021
|
+
}
|
|
29022
|
+
if (bpc === 1 || bpc === 2 || bpc === 4) {
|
|
29023
|
+
const out = new Uint8Array(count);
|
|
29024
|
+
const max = (1 << bpc) - 1;
|
|
29025
|
+
let bitPos = 0;
|
|
29026
|
+
for (let i = 0; i < count; i++) {
|
|
29027
|
+
const byteIdx = bitPos >>> 3;
|
|
29028
|
+
if (byteIdx >= data.length) return void 0;
|
|
29029
|
+
const shift = 8 - (bitPos & 7) - bpc;
|
|
29030
|
+
const val = data[byteIdx] >> shift & max;
|
|
29031
|
+
out[i] = Math.round(val / max * 255);
|
|
29032
|
+
bitPos += bpc;
|
|
29033
|
+
}
|
|
29034
|
+
return out;
|
|
29035
|
+
}
|
|
29036
|
+
}
|
|
29037
|
+
/**
|
|
29038
|
+
* Read packed single-component rows (e.g. indexed indices) honouring PDF
|
|
29039
|
+
* row-byte alignment (each scanline starts on a byte boundary).
|
|
29040
|
+
* @internal
|
|
29041
|
+
*/
|
|
29042
|
+
function readPackedRows(data, width, height, components, bpc) {
|
|
29043
|
+
const out = new Uint8Array(width * height * components);
|
|
29044
|
+
const rowBits = width * components * bpc;
|
|
29045
|
+
const rowBytes = Math.ceil(rowBits / 8);
|
|
29046
|
+
const max = (1 << bpc) - 1;
|
|
29047
|
+
for (let row = 0; row < height; row++) {
|
|
29048
|
+
const rowStart = row * rowBytes;
|
|
29049
|
+
let bitPos = 0;
|
|
29050
|
+
for (let col = 0; col < width * components; col++) {
|
|
29051
|
+
const byteIdx = rowStart + (bitPos >>> 3);
|
|
29052
|
+
let val = 0;
|
|
29053
|
+
if (byteIdx < data.length) if (bpc === 8) val = data[byteIdx];
|
|
29054
|
+
else if (bpc === 16) val = data[byteIdx];
|
|
29055
|
+
else {
|
|
29056
|
+
const shift = 8 - (bitPos & 7) - bpc;
|
|
29057
|
+
val = data[byteIdx] >> shift & max;
|
|
29058
|
+
}
|
|
29059
|
+
out[row * width * components + col] = val;
|
|
29060
|
+
bitPos += bpc;
|
|
29061
|
+
}
|
|
29062
|
+
}
|
|
29063
|
+
return out;
|
|
29064
|
+
}
|
|
29065
|
+
/**
|
|
29066
|
+
* Resolve and decode the `/SMask` of an image to a per-pixel 8-bit alpha
|
|
29067
|
+
* buffer sized `width * height`, nearest-sampling when the mask resolution
|
|
29068
|
+
* differs from the base image. Returns `undefined` when no usable mask is
|
|
29069
|
+
* present.
|
|
29070
|
+
* @internal
|
|
29071
|
+
*/
|
|
29072
|
+
function extractSoftMask(stream, registry, width, height) {
|
|
29073
|
+
const smask = resolveStream$1(stream.dict.get("/SMask"), registry);
|
|
29074
|
+
if (!smask) return void 0;
|
|
29075
|
+
const mw = readInt(smask.dict.get("/Width"), 0);
|
|
29076
|
+
const mh = readInt(smask.dict.get("/Height"), 0);
|
|
29077
|
+
if (mw <= 0 || mh <= 0) return void 0;
|
|
29078
|
+
const mbpc = readInt(smask.dict.get("/BitsPerComponent"), 8);
|
|
29079
|
+
let maskData;
|
|
29080
|
+
const { filters } = require_streamDecode.getStreamFilters(smask.dict);
|
|
29081
|
+
if (filters.includes("DCTDecode")) {
|
|
29082
|
+
const decoded = require_bridge.decodeJpegWasm(require_streamDecode.decodeStreamData(smask));
|
|
29083
|
+
if (!decoded) return void 0;
|
|
29084
|
+
maskData = decoded.pixels;
|
|
29085
|
+
} else {
|
|
29086
|
+
const raw = require_streamDecode.decodeStreamData(smask);
|
|
29087
|
+
const samples = readComponentSamples(raw, mw * mh, mbpc);
|
|
29088
|
+
if (!samples) maskData = readPackedRows(raw, mw, mh, 1, mbpc);
|
|
29089
|
+
else maskData = samples;
|
|
29090
|
+
}
|
|
29091
|
+
if (maskData.length < mw * mh) return void 0;
|
|
29092
|
+
if (mw === width && mh === height) return maskData.subarray(0, width * height);
|
|
29093
|
+
const alpha = new Uint8Array(width * height);
|
|
29094
|
+
for (let y = 0; y < height; y++) {
|
|
29095
|
+
const sy = Math.min(mh - 1, Math.floor(y * mh / height));
|
|
29096
|
+
for (let x = 0; x < width; x++) {
|
|
29097
|
+
const sx = Math.min(mw - 1, Math.floor(x * mw / width));
|
|
29098
|
+
alpha[y * width + x] = maskData[sy * mw + sx] ?? 255;
|
|
29099
|
+
}
|
|
29100
|
+
}
|
|
29101
|
+
return alpha;
|
|
29102
|
+
}
|
|
29103
|
+
/**
|
|
29104
|
+
* Composite a base colour buffer with an alpha channel into interleaved
|
|
29105
|
+
* RGBA pixels. A grayscale base is promoted to RGB first.
|
|
29106
|
+
* @internal
|
|
29107
|
+
*/
|
|
29108
|
+
function compositeAlpha(base, channels, width, height, alpha) {
|
|
29109
|
+
const pixelCount = width * height;
|
|
29110
|
+
const out = new Uint8Array(pixelCount * 4);
|
|
29111
|
+
for (let p = 0; p < pixelCount; p++) {
|
|
29112
|
+
const di = p * 4;
|
|
29113
|
+
if (channels === 1) {
|
|
29114
|
+
const g = base[p] ?? 0;
|
|
29115
|
+
out[di] = g;
|
|
29116
|
+
out[di + 1] = g;
|
|
29117
|
+
out[di + 2] = g;
|
|
29118
|
+
} else {
|
|
29119
|
+
const si = p * channels;
|
|
29120
|
+
out[di] = base[si] ?? 0;
|
|
29121
|
+
out[di + 1] = base[si + 1] ?? 0;
|
|
29122
|
+
out[di + 2] = base[si + 2] ?? 0;
|
|
29123
|
+
}
|
|
29124
|
+
out[di + 3] = alpha[p] ?? 255;
|
|
29125
|
+
}
|
|
29126
|
+
return out;
|
|
29127
|
+
}
|
|
29128
|
+
/**
|
|
29129
|
+
* Resolve a `/ColorSpace` entry into a canonical name, component count, and
|
|
29130
|
+
* (for indexed spaces) the palette bytes.
|
|
29131
|
+
* @internal
|
|
29132
|
+
*/
|
|
29133
|
+
function resolveColorSpace(entry, registry) {
|
|
29134
|
+
const resolved = deref(entry, registry);
|
|
29135
|
+
if (resolved instanceof require_pdfObjects.PdfName) return nameToColorSpace(resolved.value);
|
|
29136
|
+
if (resolved instanceof require_pdfObjects.PdfArray) {
|
|
29137
|
+
const first = deref(resolved.items[0], registry);
|
|
29138
|
+
if (first instanceof require_pdfObjects.PdfName) {
|
|
29139
|
+
const head = stripSlash(first.value);
|
|
29140
|
+
if (head === "ICCBased") {
|
|
29141
|
+
const profile = resolveStream$1(resolved.items[1], registry);
|
|
29142
|
+
const n = profile ? readInt(profile.dict.get("/N"), 3) : 3;
|
|
29143
|
+
if (n === 1) return {
|
|
29144
|
+
name: "DeviceGray",
|
|
29145
|
+
components: 1
|
|
29146
|
+
};
|
|
29147
|
+
if (n === 4) return {
|
|
29148
|
+
name: "DeviceCMYK",
|
|
29149
|
+
components: 4
|
|
29150
|
+
};
|
|
29151
|
+
return {
|
|
29152
|
+
name: "DeviceRGB",
|
|
29153
|
+
components: 3
|
|
29154
|
+
};
|
|
29155
|
+
}
|
|
29156
|
+
if (head === "Indexed" || head === "I") return {
|
|
29157
|
+
name: "Indexed",
|
|
29158
|
+
components: 1,
|
|
29159
|
+
palette: readIndexedPalette(resolved, registry)
|
|
29160
|
+
};
|
|
29161
|
+
if (head === "CalRGB" || head === "Lab") return {
|
|
29162
|
+
name: head,
|
|
29163
|
+
components: 3
|
|
29164
|
+
};
|
|
29165
|
+
if (head === "CalGray") return {
|
|
29166
|
+
name: "CalGray",
|
|
29167
|
+
components: 1
|
|
29168
|
+
};
|
|
29169
|
+
return nameToColorSpace(first.value);
|
|
29170
|
+
}
|
|
29171
|
+
}
|
|
29172
|
+
return {
|
|
29173
|
+
name: "DeviceRGB",
|
|
29174
|
+
components: 3
|
|
29175
|
+
};
|
|
29176
|
+
}
|
|
29177
|
+
/** Map a colour-space name to its canonical form + component count. @internal */
|
|
29178
|
+
function nameToColorSpace(rawName) {
|
|
29179
|
+
switch (stripSlash(rawName)) {
|
|
29180
|
+
case "DeviceGray":
|
|
29181
|
+
case "G":
|
|
29182
|
+
case "CalGray": return {
|
|
29183
|
+
name: "DeviceGray",
|
|
29184
|
+
components: 1
|
|
29185
|
+
};
|
|
29186
|
+
case "DeviceCMYK":
|
|
29187
|
+
case "CMYK": return {
|
|
29188
|
+
name: "DeviceCMYK",
|
|
29189
|
+
components: 4
|
|
29190
|
+
};
|
|
29191
|
+
case "DeviceRGB":
|
|
29192
|
+
case "RGB":
|
|
29193
|
+
case "CalRGB": return {
|
|
29194
|
+
name: "DeviceRGB",
|
|
29195
|
+
components: 3
|
|
29196
|
+
};
|
|
29197
|
+
default: return {
|
|
29198
|
+
name: "DeviceRGB",
|
|
29199
|
+
components: 3
|
|
29200
|
+
};
|
|
29201
|
+
}
|
|
29202
|
+
}
|
|
29203
|
+
/**
|
|
29204
|
+
* Read the lookup table (palette) from an `[/Indexed base hival lookup]`
|
|
29205
|
+
* colour-space array. The lookup table may be a hex/literal string or a
|
|
29206
|
+
* stream.
|
|
29207
|
+
* @internal
|
|
29208
|
+
*/
|
|
29209
|
+
function readIndexedPalette(arr, registry) {
|
|
29210
|
+
const lookup = deref(arr.items[3], registry);
|
|
29211
|
+
if (lookup instanceof require_pdfObjects.PdfString) {
|
|
29212
|
+
const value = lookup.value;
|
|
29213
|
+
const bytes = new Uint8Array(value.length);
|
|
29214
|
+
for (let i = 0; i < value.length; i++) bytes[i] = value.charCodeAt(i) & 255;
|
|
29215
|
+
return bytes;
|
|
29216
|
+
}
|
|
29217
|
+
if (lookup instanceof require_pdfObjects.PdfStream) return require_streamDecode.decodeStreamData(lookup);
|
|
29218
|
+
}
|
|
29219
|
+
/** Resolve a value to its target object, following one indirect reference. @internal */
|
|
29220
|
+
function deref(obj, registry) {
|
|
29221
|
+
if (obj instanceof require_pdfObjects.PdfRef) return registry.resolve(obj);
|
|
29222
|
+
return obj;
|
|
29223
|
+
}
|
|
29224
|
+
/** Resolve a value (possibly a ref) to a `PdfDict`, or `undefined`. @internal */
|
|
29225
|
+
function resolveDict$1(obj, registry) {
|
|
29226
|
+
const resolved = deref(obj, registry);
|
|
29227
|
+
return resolved instanceof require_pdfObjects.PdfDict ? resolved : void 0;
|
|
29228
|
+
}
|
|
29229
|
+
/** Resolve a value (possibly a ref) to a `PdfStream`, or `undefined`. @internal */
|
|
29230
|
+
function resolveStream$1(obj, registry) {
|
|
29231
|
+
const resolved = deref(obj, registry);
|
|
29232
|
+
return resolved instanceof require_pdfObjects.PdfStream ? resolved : void 0;
|
|
29233
|
+
}
|
|
29234
|
+
/** Read an integer from a PDF object, or fall back. @internal */
|
|
29235
|
+
function readInt(obj, fallback) {
|
|
29236
|
+
return obj instanceof require_pdfObjects.PdfNumber ? Math.round(obj.value) : fallback;
|
|
29237
|
+
}
|
|
29238
|
+
/** Strip a single leading slash from a PDF name string. @internal */
|
|
29239
|
+
function stripSlash(s) {
|
|
29240
|
+
return s.startsWith("/") ? s.slice(1) : s;
|
|
29241
|
+
}
|
|
29242
|
+
//#endregion
|
|
29243
|
+
//#region src/render/fontExtract.ts
|
|
29244
|
+
/**
|
|
29245
|
+
* Resolve an object through the registry if it is an indirect reference,
|
|
29246
|
+
* otherwise return it unchanged.
|
|
29247
|
+
*/
|
|
29248
|
+
function resolve(obj, registry) {
|
|
29249
|
+
if (obj && obj.kind === "ref") return registry.resolve(obj);
|
|
29250
|
+
return obj;
|
|
29251
|
+
}
|
|
29252
|
+
/** Resolve `dict[key]` and return it only if it is a dictionary. */
|
|
29253
|
+
function resolveDict(dict, key, registry) {
|
|
29254
|
+
const v = resolve(dict.get(key), registry);
|
|
29255
|
+
return v && v.kind === "dict" ? v : void 0;
|
|
29256
|
+
}
|
|
29257
|
+
/** Resolve `dict[key]` and return it only if it is a stream. */
|
|
29258
|
+
function resolveStream(dict, key, registry) {
|
|
29259
|
+
const v = resolve(dict.get(key), registry);
|
|
29260
|
+
return v && v.kind === "stream" ? v : void 0;
|
|
29261
|
+
}
|
|
29262
|
+
/** Read a `/Name` value with the leading slash stripped, if present. */
|
|
29263
|
+
function nameValue(obj, registry) {
|
|
29264
|
+
const v = resolve(obj, registry);
|
|
29265
|
+
if (v && v.kind === "name") {
|
|
29266
|
+
const raw = v.value;
|
|
29267
|
+
return raw.startsWith("/") ? raw.slice(1) : raw;
|
|
29268
|
+
}
|
|
29269
|
+
}
|
|
29270
|
+
/**
|
|
29271
|
+
* Locate the `/FontDescriptor` for a font dictionary.
|
|
29272
|
+
*
|
|
29273
|
+
* For simple fonts the descriptor is a direct child. For composite
|
|
29274
|
+
* (Type0) fonts it lives on the first `/DescendantFonts` entry.
|
|
29275
|
+
*/
|
|
29276
|
+
function findFontDescriptor(fontDict, registry) {
|
|
29277
|
+
const direct = resolveDict(fontDict, "/FontDescriptor", registry);
|
|
29278
|
+
if (direct) return direct;
|
|
29279
|
+
if (nameValue(fontDict.get("/Subtype"), registry) === "Type0") {
|
|
29280
|
+
const descendants = resolve(fontDict.get("/DescendantFonts"), registry);
|
|
29281
|
+
if (descendants && descendants.kind === "array") {
|
|
29282
|
+
const first = descendants.items[0];
|
|
29283
|
+
const cidFont = resolve(first, registry);
|
|
29284
|
+
if (cidFont && cidFont.kind === "dict") return resolveDict(cidFont, "/FontDescriptor", registry);
|
|
29285
|
+
}
|
|
29286
|
+
}
|
|
29287
|
+
}
|
|
29288
|
+
/**
|
|
29289
|
+
* Pull the embedded font program (if any) out of a `/FontDescriptor`.
|
|
29290
|
+
*
|
|
29291
|
+
* Returns the decoded bytes and the detected format, or `undefined` when
|
|
29292
|
+
* the descriptor carries no embedded file (e.g. a standard font).
|
|
29293
|
+
*/
|
|
29294
|
+
function extractFromDescriptor(descriptor, registry) {
|
|
29295
|
+
const fontFile = resolveStream(descriptor, "/FontFile", registry);
|
|
29296
|
+
if (fontFile) return {
|
|
29297
|
+
format: "type1",
|
|
29298
|
+
data: require_streamDecode.decodeStreamData(fontFile)
|
|
29299
|
+
};
|
|
29300
|
+
const fontFile2 = resolveStream(descriptor, "/FontFile2", registry);
|
|
29301
|
+
if (fontFile2) return {
|
|
29302
|
+
format: "truetype",
|
|
29303
|
+
data: require_streamDecode.decodeStreamData(fontFile2)
|
|
29304
|
+
};
|
|
29305
|
+
const fontFile3 = resolveStream(descriptor, "/FontFile3", registry);
|
|
29306
|
+
if (fontFile3) return {
|
|
29307
|
+
format: nameValue(fontFile3.dict.get("/Subtype"), registry) === "OpenType" ? "opentype" : "cff",
|
|
29308
|
+
data: require_streamDecode.decodeStreamData(fontFile3)
|
|
29309
|
+
};
|
|
29310
|
+
}
|
|
29311
|
+
/**
|
|
29312
|
+
* Determine whether a base-font name carries a subset tag.
|
|
29313
|
+
*
|
|
29314
|
+
* A subset tag is exactly six uppercase ASCII letters followed by `+`,
|
|
29315
|
+
* e.g. `ABCDEF+Helvetica`.
|
|
29316
|
+
*/
|
|
29317
|
+
function hasSubsetTag(baseFont) {
|
|
29318
|
+
if (baseFont.length < 7 || baseFont[6] !== "+") return false;
|
|
29319
|
+
for (let i = 0; i < 6; i++) {
|
|
29320
|
+
const code = baseFont.charCodeAt(i);
|
|
29321
|
+
if (code < 65 || code > 90) return false;
|
|
29322
|
+
}
|
|
29323
|
+
return true;
|
|
29324
|
+
}
|
|
29325
|
+
/**
|
|
29326
|
+
* Extract every embedded font program referenced by a page.
|
|
29327
|
+
*
|
|
29328
|
+
* Walks the page's original `/Resources` → `/Font` dictionary. For each
|
|
29329
|
+
* font that carries an embedded file (`/FontFile`, `/FontFile2`, or
|
|
29330
|
+
* `/FontFile3`) — directly or via a Type0 descendant — the standalone,
|
|
29331
|
+
* filter-decoded font bytes are returned. Standard-14 fonts with no
|
|
29332
|
+
* embedded file are omitted, and malformed entries are skipped silently.
|
|
29333
|
+
*
|
|
29334
|
+
* @param page The page to scan (typically from a loaded PDF).
|
|
29335
|
+
* @returns One {@link ExtractedFont} per embedded font program.
|
|
29336
|
+
*
|
|
29337
|
+
* @example
|
|
29338
|
+
* ```ts
|
|
29339
|
+
* const loaded = await PdfDocument.load(bytes);
|
|
29340
|
+
* const fonts = extractFonts(loaded.getPage(0));
|
|
29341
|
+
* for (const f of fonts) {
|
|
29342
|
+
* console.log(f.resourceName, f.baseFont, f.format, f.data.length, f.subset);
|
|
29343
|
+
* }
|
|
29344
|
+
* ```
|
|
29345
|
+
*/
|
|
29346
|
+
function extractFonts(page) {
|
|
29347
|
+
const resources = page.getOriginalResources();
|
|
29348
|
+
if (!resources) return [];
|
|
29349
|
+
const registry = page.getRegistry();
|
|
29350
|
+
const fontDict = resolveDict(resources, "/Font", registry);
|
|
29351
|
+
if (!fontDict) return [];
|
|
29352
|
+
const result = [];
|
|
29353
|
+
for (const [key, value] of fontDict) {
|
|
29354
|
+
const dict = resolve(value, registry);
|
|
29355
|
+
if (!dict || dict.kind !== "dict") continue;
|
|
29356
|
+
const descriptor = findFontDescriptor(dict, registry);
|
|
29357
|
+
if (!descriptor) continue;
|
|
29358
|
+
let extracted;
|
|
29359
|
+
try {
|
|
29360
|
+
extracted = extractFromDescriptor(descriptor, registry);
|
|
29361
|
+
} catch {
|
|
29362
|
+
continue;
|
|
29363
|
+
}
|
|
29364
|
+
if (!extracted) continue;
|
|
29365
|
+
const baseFont = nameValue(dict.get("/BaseFont"), registry) ?? "";
|
|
29366
|
+
const resourceName = key.startsWith("/") ? key.slice(1) : key;
|
|
29367
|
+
result.push({
|
|
29368
|
+
resourceName,
|
|
29369
|
+
baseFont,
|
|
29370
|
+
format: extracted.format,
|
|
29371
|
+
data: extracted.data,
|
|
29372
|
+
subset: hasSubsetTag(baseFont)
|
|
29373
|
+
});
|
|
29374
|
+
}
|
|
29375
|
+
return result;
|
|
29376
|
+
}
|
|
29377
|
+
//#endregion
|
|
29378
|
+
//#region src/render/diff.ts
|
|
29379
|
+
/**
|
|
29380
|
+
* @module render/diff
|
|
29381
|
+
*
|
|
29382
|
+
* Visual regression for PDFs: rasterize two pages (or compare two raster
|
|
29383
|
+
* images) and report a per-pixel difference count, a difference heatmap, and a
|
|
29384
|
+
* structural-similarity (SSIM) score over 8×8 luminance blocks.
|
|
29385
|
+
*
|
|
29386
|
+
* @packageDocumentation
|
|
29387
|
+
*/
|
|
29388
|
+
const C1 = (.01 * 255) ** 2;
|
|
29389
|
+
const C2 = (.03 * 255) ** 2;
|
|
29390
|
+
function luma(data, i) {
|
|
29391
|
+
return .299 * data[i] + .587 * data[i + 1] + .114 * data[i + 2];
|
|
29392
|
+
}
|
|
29393
|
+
/** Compare two equally-sized raster images. */
|
|
29394
|
+
function compareImages(a, b, opts = {}) {
|
|
29395
|
+
if (a.width !== b.width || a.height !== b.height) throw new Error(`compareImages: size mismatch (${a.width}×${a.height} vs ${b.width}×${b.height}); render both at the same scale`);
|
|
29396
|
+
const W = a.width;
|
|
29397
|
+
const H = a.height;
|
|
29398
|
+
const threshold = opts.threshold ?? 16;
|
|
29399
|
+
const heatmap = new Uint8Array(W * H * 4);
|
|
29400
|
+
let changed = 0;
|
|
29401
|
+
for (let p = 0; p < W * H; p++) {
|
|
29402
|
+
const i = p * 4;
|
|
29403
|
+
const dr = Math.abs(a.data[i] - b.data[i]);
|
|
29404
|
+
const dg = Math.abs(a.data[i + 1] - b.data[i + 1]);
|
|
29405
|
+
const db = Math.abs(a.data[i + 2] - b.data[i + 2]);
|
|
29406
|
+
if (Math.max(dr, dg, db) > threshold) {
|
|
29407
|
+
changed++;
|
|
29408
|
+
heatmap[i] = 255;
|
|
29409
|
+
heatmap[i + 1] = 40;
|
|
29410
|
+
heatmap[i + 2] = 40;
|
|
29411
|
+
heatmap[i + 3] = 255;
|
|
29412
|
+
} else {
|
|
29413
|
+
const g = Math.round(luma(a.data, i) * .4 + 153);
|
|
29414
|
+
heatmap[i] = g;
|
|
29415
|
+
heatmap[i + 1] = g;
|
|
29416
|
+
heatmap[i + 2] = g;
|
|
29417
|
+
heatmap[i + 3] = 255;
|
|
29418
|
+
}
|
|
29419
|
+
}
|
|
29420
|
+
return {
|
|
29421
|
+
width: W,
|
|
29422
|
+
height: H,
|
|
29423
|
+
changedPixels: changed,
|
|
29424
|
+
changedRatio: changed / (W * H),
|
|
29425
|
+
ssim: meanSsim(a, b, W, H),
|
|
29426
|
+
heatmap
|
|
29427
|
+
};
|
|
29428
|
+
}
|
|
29429
|
+
/** Mean SSIM over non-overlapping 8×8 luminance blocks. */
|
|
29430
|
+
function meanSsim(a, b, W, H) {
|
|
29431
|
+
const block = 8;
|
|
29432
|
+
let total = 0;
|
|
29433
|
+
let count = 0;
|
|
29434
|
+
for (let by = 0; by < H; by += block) for (let bx = 0; bx < W; bx += block) {
|
|
29435
|
+
const bw = Math.min(block, W - bx);
|
|
29436
|
+
const bh = Math.min(block, H - by);
|
|
29437
|
+
const n = bw * bh;
|
|
29438
|
+
let sa = 0;
|
|
29439
|
+
let sb = 0;
|
|
29440
|
+
let saa = 0;
|
|
29441
|
+
let sbb = 0;
|
|
29442
|
+
let sab = 0;
|
|
29443
|
+
for (let y = 0; y < bh; y++) for (let x = 0; x < bw; x++) {
|
|
29444
|
+
const i = ((by + y) * W + (bx + x)) * 4;
|
|
29445
|
+
const la = luma(a.data, i);
|
|
29446
|
+
const lb = luma(b.data, i);
|
|
29447
|
+
sa += la;
|
|
29448
|
+
sb += lb;
|
|
29449
|
+
saa += la * la;
|
|
29450
|
+
sbb += lb * lb;
|
|
29451
|
+
sab += la * lb;
|
|
29452
|
+
}
|
|
29453
|
+
const ma = sa / n;
|
|
29454
|
+
const mb = sb / n;
|
|
29455
|
+
const va = saa / n - ma * ma;
|
|
29456
|
+
const vb = sbb / n - mb * mb;
|
|
29457
|
+
const cov = sab / n - ma * mb;
|
|
29458
|
+
const ssim = (2 * ma * mb + C1) * (2 * cov + C2) / ((ma * ma + mb * mb + C1) * (va + vb + C2));
|
|
29459
|
+
total += ssim;
|
|
29460
|
+
count++;
|
|
29461
|
+
}
|
|
29462
|
+
return count ? total / count : 1;
|
|
29463
|
+
}
|
|
29464
|
+
/** Rasterize two pages at the same scale and compare them. */
|
|
29465
|
+
async function comparePages(pageA, pageB, opts = {}) {
|
|
29466
|
+
return compareImages(rasterize(interpretPage(pageA), opts), rasterize(interpretPage(pageB), opts), opts);
|
|
29467
|
+
}
|
|
29468
|
+
//#endregion
|
|
29469
|
+
//#region src/render/ocr.ts
|
|
29470
|
+
/**
|
|
29471
|
+
* Escape a string for use inside a PDF literal string `(…)`.
|
|
29472
|
+
* Backslash and both parentheses must be escaped so the content stream
|
|
29473
|
+
* remains well-formed.
|
|
29474
|
+
*/
|
|
29475
|
+
function escapePdfLiteral(text) {
|
|
29476
|
+
return text.replaceAll("\\", "\\\\").replaceAll("(", "\\(").replaceAll(")", "\\)");
|
|
29477
|
+
}
|
|
29478
|
+
/**
|
|
29479
|
+
* Format a number for PDF output: at most 4 decimal places, trailing zeros
|
|
29480
|
+
* stripped, never `-0`.
|
|
29481
|
+
*/
|
|
29482
|
+
function num$1(value) {
|
|
29483
|
+
if (!Number.isFinite(value)) return "0";
|
|
29484
|
+
if (Number.isInteger(value)) return value.toString();
|
|
29485
|
+
const s = value.toFixed(4).replace(/\.?0+$/, "");
|
|
29486
|
+
return s === "-0" ? "0" : s;
|
|
29487
|
+
}
|
|
29488
|
+
/**
|
|
29489
|
+
* Build a standard Helvetica (Type 1) font dictionary, register it in the
|
|
29490
|
+
* page's object registry, and attach it to the page under `resourceName`.
|
|
29491
|
+
*
|
|
29492
|
+
* This lets {@link applyOcr} guarantee a usable font resource exists for the
|
|
29493
|
+
* invisible overlay without needing a reference to the owning document.
|
|
29494
|
+
*/
|
|
29495
|
+
function ensureOverlayFont(page, resourceName) {
|
|
29496
|
+
const registry = page.getRegistry();
|
|
29497
|
+
const fontDict = new require_pdfObjects.PdfDict();
|
|
29498
|
+
fontDict.set("/Type", require_pdfObjects.PdfName.of("Font"));
|
|
29499
|
+
fontDict.set("/Subtype", require_pdfObjects.PdfName.of("Type1"));
|
|
29500
|
+
fontDict.set("/BaseFont", require_pdfObjects.PdfName.of("Helvetica"));
|
|
29501
|
+
fontDict.set("/Encoding", require_pdfObjects.PdfName.of("WinAnsiEncoding"));
|
|
29502
|
+
const ref = registry.register(fontDict);
|
|
29503
|
+
page.registerFont(resourceName, ref);
|
|
29504
|
+
}
|
|
29505
|
+
/**
|
|
29506
|
+
* Run OCR on a page and append the recognized text as an invisible,
|
|
29507
|
+
* selectable/searchable overlay.
|
|
29508
|
+
*
|
|
29509
|
+
* The page is rasterized to RGBA at `opts.dpi`, passed to `engine.recognize`,
|
|
29510
|
+
* and each returned word is written back as a content-stream fragment:
|
|
29511
|
+
*
|
|
29512
|
+
* ```text
|
|
29513
|
+
* BT 3 Tr /<font> <size> Tf <x> <y> Td (<escaped text>) Tj ET
|
|
29514
|
+
* ```
|
|
29515
|
+
*
|
|
29516
|
+
* where `<size> ≈ word.height` and `<x> <y>` is the word's lower-left corner
|
|
29517
|
+
* in page space. Render mode `3` makes the text invisible (no fill, no
|
|
29518
|
+
* stroke) while keeping it part of the document's text content.
|
|
29519
|
+
*
|
|
29520
|
+
* @param page The page to OCR and annotate.
|
|
29521
|
+
* @param engine The OCR backend.
|
|
29522
|
+
* @param opts Optional DPI and font-resource overrides.
|
|
29523
|
+
* @returns The words returned by the engine (unmodified).
|
|
29524
|
+
*/
|
|
29525
|
+
async function applyOcr(page, engine, opts) {
|
|
29526
|
+
const dpi = opts?.dpi ?? 150;
|
|
29527
|
+
const fontName = opts?.fontResourceName ?? "OCRFont";
|
|
29528
|
+
const raster = rasterize(interpretPage(page), { dpi });
|
|
29529
|
+
const words = await engine.recognize(raster.data, raster.width, raster.height);
|
|
29530
|
+
if (words.length === 0) return words;
|
|
29531
|
+
ensureOverlayFont(page, fontName);
|
|
29532
|
+
let fragment = "";
|
|
29533
|
+
for (const word of words) {
|
|
29534
|
+
const size = word.height > 0 ? word.height : 1;
|
|
29535
|
+
const escaped = escapePdfLiteral(word.text);
|
|
29536
|
+
fragment += `BT 3 Tr /${fontName} ${num$1(size)} Tf ${num$1(word.x)} ${num$1(word.y)} Td (${escaped}) Tj ET\n`;
|
|
29537
|
+
}
|
|
29538
|
+
page.pushOperators(fragment);
|
|
29539
|
+
return words;
|
|
29540
|
+
}
|
|
29541
|
+
//#endregion
|
|
29542
|
+
//#region src/render/redactContent.ts
|
|
29543
|
+
/** Build a {@link BBox} from a set of page-space corner points. */
|
|
29544
|
+
function bboxOfPoints(points) {
|
|
29545
|
+
let minX = Infinity;
|
|
29546
|
+
let minY = Infinity;
|
|
29547
|
+
let maxX = -Infinity;
|
|
29548
|
+
let maxY = -Infinity;
|
|
29549
|
+
for (const [x, y] of points) {
|
|
29550
|
+
if (x < minX) minX = x;
|
|
29551
|
+
if (y < minY) minY = y;
|
|
29552
|
+
if (x > maxX) maxX = x;
|
|
29553
|
+
if (y > maxY) maxY = y;
|
|
29554
|
+
}
|
|
29555
|
+
return {
|
|
29556
|
+
minX,
|
|
29557
|
+
minY,
|
|
29558
|
+
maxX,
|
|
29559
|
+
maxY
|
|
29560
|
+
};
|
|
29561
|
+
}
|
|
29562
|
+
/** Whether a page-space bounding box intersects a redaction rect. */
|
|
29563
|
+
function intersects(box, rect) {
|
|
29564
|
+
const rMinX = rect.x;
|
|
29565
|
+
const rMinY = rect.y;
|
|
29566
|
+
const rMaxX = rect.x + rect.width;
|
|
29567
|
+
const rMaxY = rect.y + rect.height;
|
|
29568
|
+
return box.minX <= rMaxX && box.maxX >= rMinX && box.minY <= rMaxY && box.maxY >= rMinY;
|
|
29569
|
+
}
|
|
29570
|
+
/** Whether a bounding box intersects *any* of the redaction rects. */
|
|
29571
|
+
function intersectsAny(box, rects) {
|
|
29572
|
+
for (const rect of rects) {
|
|
29573
|
+
if (rect.width <= 0 || rect.height <= 0) continue;
|
|
29574
|
+
if (intersects(box, rect)) return true;
|
|
29575
|
+
}
|
|
29576
|
+
return false;
|
|
29577
|
+
}
|
|
29578
|
+
function num(op) {
|
|
29579
|
+
return typeof op === "number" ? op : 0;
|
|
29580
|
+
}
|
|
29581
|
+
/** Length of the visible characters a text-show operand contributes. */
|
|
29582
|
+
function operandTextLength(op) {
|
|
29583
|
+
if (typeof op === "string") return op.length;
|
|
29584
|
+
if (Array.isArray(op)) {
|
|
29585
|
+
let total = 0;
|
|
29586
|
+
for (const el of op) if (typeof el === "string") total += el.length;
|
|
29587
|
+
return total;
|
|
29588
|
+
}
|
|
29589
|
+
return 0;
|
|
29590
|
+
}
|
|
29591
|
+
const AVG_GLYPH_WIDTH_EM = .5;
|
|
29592
|
+
const ASCENT_EM = .85;
|
|
29593
|
+
const DESCENT_EM = -.2;
|
|
29594
|
+
/**
|
|
29595
|
+
* Compute the page-space bounding box of a text run shown at the current
|
|
29596
|
+
* text-rendering matrix. The run is approximated as a rectangle spanning
|
|
29597
|
+
* `[0, advance]` horizontally and `[descent, ascent]` vertically in text
|
|
29598
|
+
* space (em units, before the font-size scale baked into `trm`).
|
|
29599
|
+
*/
|
|
29600
|
+
function textRunBBox(trm, glyphCount) {
|
|
29601
|
+
const advance = Math.max(glyphCount, 1) * AVG_GLYPH_WIDTH_EM;
|
|
29602
|
+
return bboxOfPoints([
|
|
29603
|
+
applyToPoint(trm, 0, DESCENT_EM),
|
|
29604
|
+
applyToPoint(trm, advance, DESCENT_EM),
|
|
29605
|
+
applyToPoint(trm, advance, ASCENT_EM),
|
|
29606
|
+
applyToPoint(trm, 0, ASCENT_EM)
|
|
29607
|
+
]);
|
|
29608
|
+
}
|
|
29609
|
+
/**
|
|
29610
|
+
* Compute the page-space bounding box of an image placed by `Do`. An image
|
|
29611
|
+
* XObject is painted into the unit square `[0,1]×[0,1]` transformed by the
|
|
29612
|
+
* current CTM.
|
|
29613
|
+
*/
|
|
29614
|
+
function imageBBox(ctm) {
|
|
29615
|
+
return bboxOfPoints([
|
|
29616
|
+
applyToPoint(ctm, 0, 0),
|
|
29617
|
+
applyToPoint(ctm, 1, 0),
|
|
29618
|
+
applyToPoint(ctm, 1, 1),
|
|
29619
|
+
applyToPoint(ctm, 0, 1)
|
|
29620
|
+
]);
|
|
29621
|
+
}
|
|
29622
|
+
/**
|
|
29623
|
+
* Serialize a string operand as a PDF hex string `<…>`.
|
|
29624
|
+
*
|
|
29625
|
+
* Hex strings are byte-exact and contain only ASCII, so they survive the
|
|
29626
|
+
* UTF-8 encoding applied when the content stream is written. The parser
|
|
29627
|
+
* decodes both `(…)` and `<…>` strings to a JS string of latin-1 char codes,
|
|
29628
|
+
* so re-emitting as hex preserves the original bytes regardless of the
|
|
29629
|
+
* original literal form (important for multi-byte / CID-encoded text).
|
|
29630
|
+
*/
|
|
29631
|
+
function serializeString(s) {
|
|
29632
|
+
let out = "<";
|
|
29633
|
+
for (let i = 0; i < s.length; i++) {
|
|
29634
|
+
const byte = s.charCodeAt(i) & 255;
|
|
29635
|
+
out += byte.toString(16).padStart(2, "0");
|
|
29636
|
+
}
|
|
29637
|
+
return out + ">";
|
|
29638
|
+
}
|
|
29639
|
+
/** Format a number the way PDF content streams expect (no trailing zeros). */
|
|
29640
|
+
function formatNumber(value) {
|
|
29641
|
+
if (Number.isInteger(value)) return value.toString();
|
|
29642
|
+
const s = value.toFixed(6).replace(/\.?0+$/, "");
|
|
29643
|
+
return s === "-0" ? "0" : s;
|
|
29644
|
+
}
|
|
29645
|
+
/** Serialize a single operand to its content-stream textual form. */
|
|
29646
|
+
function serializeOperand(op) {
|
|
29647
|
+
if (op === null) return "null";
|
|
29648
|
+
if (typeof op === "boolean") return op ? "true" : "false";
|
|
29649
|
+
if (typeof op === "number") return formatNumber(op);
|
|
29650
|
+
if (typeof op === "string") return serializeString(op);
|
|
29651
|
+
if (op instanceof require_pdfObjects.PdfName) return op.value;
|
|
29652
|
+
if (Array.isArray(op)) return "[" + op.map(serializeOperand).join(" ") + "]";
|
|
29653
|
+
return "";
|
|
29654
|
+
}
|
|
29655
|
+
/** Serialize one operator (operands followed by the operator keyword). */
|
|
29656
|
+
function serializeOperator(op) {
|
|
29657
|
+
if (op.operands.length === 0) return op.operator;
|
|
29658
|
+
return `${op.operands.map(serializeOperand).join(" ")} ${op.operator}`;
|
|
29659
|
+
}
|
|
29660
|
+
function initialState() {
|
|
29661
|
+
return {
|
|
29662
|
+
ctm: identity(),
|
|
29663
|
+
fontSize: 0,
|
|
29664
|
+
hScale: 1,
|
|
29665
|
+
rise: 0,
|
|
29666
|
+
leading: 0
|
|
29667
|
+
};
|
|
29668
|
+
}
|
|
29669
|
+
/**
|
|
29670
|
+
* Build the text-rendering matrix used to map text-space (em) coordinates to
|
|
29671
|
+
* page space, matching the interpreter:
|
|
29672
|
+
* `Trm = [fontSize·hScale, 0, 0, fontSize, 0, rise] × Tm × CTM`.
|
|
29673
|
+
*/
|
|
29674
|
+
function textRenderingMatrix(state, tm) {
|
|
29675
|
+
return multiply(multiply([
|
|
29676
|
+
state.fontSize * state.hScale,
|
|
29677
|
+
0,
|
|
29678
|
+
0,
|
|
29679
|
+
state.fontSize,
|
|
29680
|
+
0,
|
|
29681
|
+
state.rise
|
|
29682
|
+
], tm), state.ctm);
|
|
29683
|
+
}
|
|
29684
|
+
/**
|
|
29685
|
+
* Walk the operators, tracking position, and return both the filtered
|
|
29686
|
+
* operator list and the removal counts.
|
|
29687
|
+
*/
|
|
29688
|
+
function filterOperators(operators, rects) {
|
|
29689
|
+
const kept = [];
|
|
29690
|
+
let removedText = 0;
|
|
29691
|
+
let removedImages = 0;
|
|
29692
|
+
const stack = [];
|
|
29693
|
+
let state = initialState();
|
|
29694
|
+
let tm = identity();
|
|
29695
|
+
let tlm = identity();
|
|
29696
|
+
for (const op of operators) {
|
|
29697
|
+
const { operator, operands } = op;
|
|
29698
|
+
let drop = false;
|
|
29699
|
+
switch (operator) {
|
|
29700
|
+
case "q":
|
|
29701
|
+
stack.push({ ...state });
|
|
29702
|
+
break;
|
|
29703
|
+
case "Q":
|
|
29704
|
+
if (stack.length > 0) state = stack.pop();
|
|
29705
|
+
break;
|
|
29706
|
+
case "cm":
|
|
29707
|
+
state = {
|
|
29708
|
+
...state,
|
|
29709
|
+
ctm: multiply([
|
|
29710
|
+
num(operands[0]),
|
|
29711
|
+
num(operands[1]),
|
|
29712
|
+
num(operands[2]),
|
|
29713
|
+
num(operands[3]),
|
|
29714
|
+
num(operands[4]),
|
|
29715
|
+
num(operands[5])
|
|
29716
|
+
], state.ctm)
|
|
29717
|
+
};
|
|
29718
|
+
break;
|
|
29719
|
+
case "BT":
|
|
29720
|
+
tm = identity();
|
|
29721
|
+
tlm = identity();
|
|
29722
|
+
break;
|
|
29723
|
+
case "ET": break;
|
|
29724
|
+
case "Tf":
|
|
29725
|
+
state = {
|
|
29726
|
+
...state,
|
|
29727
|
+
fontSize: num(operands[1])
|
|
29728
|
+
};
|
|
29729
|
+
break;
|
|
29730
|
+
case "Tz":
|
|
29731
|
+
state = {
|
|
29732
|
+
...state,
|
|
29733
|
+
hScale: num(operands[0]) / 100
|
|
29734
|
+
};
|
|
29735
|
+
break;
|
|
29736
|
+
case "Ts":
|
|
29737
|
+
state = {
|
|
29738
|
+
...state,
|
|
29739
|
+
rise: num(operands[0])
|
|
29740
|
+
};
|
|
29741
|
+
break;
|
|
29742
|
+
case "TL":
|
|
29743
|
+
state = {
|
|
29744
|
+
...state,
|
|
29745
|
+
leading: num(operands[0])
|
|
29746
|
+
};
|
|
29747
|
+
break;
|
|
29748
|
+
case "Td":
|
|
29749
|
+
tlm = multiply(translation(num(operands[0]), num(operands[1])), tlm);
|
|
29750
|
+
tm = tlm;
|
|
29751
|
+
break;
|
|
29752
|
+
case "TD":
|
|
29753
|
+
state = {
|
|
29754
|
+
...state,
|
|
29755
|
+
leading: -num(operands[1])
|
|
29756
|
+
};
|
|
29757
|
+
tlm = multiply(translation(num(operands[0]), num(operands[1])), tlm);
|
|
29758
|
+
tm = tlm;
|
|
29759
|
+
break;
|
|
29760
|
+
case "Tm":
|
|
29761
|
+
tm = [
|
|
29762
|
+
num(operands[0]),
|
|
29763
|
+
num(operands[1]),
|
|
29764
|
+
num(operands[2]),
|
|
29765
|
+
num(operands[3]),
|
|
29766
|
+
num(operands[4]),
|
|
29767
|
+
num(operands[5])
|
|
29768
|
+
];
|
|
29769
|
+
tlm = tm;
|
|
29770
|
+
break;
|
|
29771
|
+
case "T*":
|
|
29772
|
+
tlm = multiply(translation(0, -state.leading), tlm);
|
|
29773
|
+
tm = tlm;
|
|
29774
|
+
break;
|
|
29775
|
+
case "Tj":
|
|
29776
|
+
case "TJ":
|
|
29777
|
+
case "'":
|
|
29778
|
+
case "\"": {
|
|
29779
|
+
if (operator === "'" || operator === "\"") {
|
|
29780
|
+
tlm = multiply(translation(0, -state.leading), tlm);
|
|
29781
|
+
tm = tlm;
|
|
29782
|
+
}
|
|
29783
|
+
if (operator === "\"") state = { ...state };
|
|
29784
|
+
const glyphCount = operandTextLength(operator === "\"" ? operands[2] : operator === "'" ? operands[0] : operands[0]);
|
|
29785
|
+
if (intersectsAny(textRunBBox(textRenderingMatrix(state, tm), glyphCount), rects)) {
|
|
29786
|
+
drop = true;
|
|
29787
|
+
removedText++;
|
|
29788
|
+
}
|
|
29789
|
+
break;
|
|
29790
|
+
}
|
|
29791
|
+
case "Do":
|
|
29792
|
+
if (intersectsAny(imageBBox(state.ctm), rects)) {
|
|
29793
|
+
drop = true;
|
|
29794
|
+
removedImages++;
|
|
29795
|
+
}
|
|
29796
|
+
break;
|
|
29797
|
+
default: break;
|
|
29798
|
+
}
|
|
29799
|
+
if (!drop) kept.push(op);
|
|
29800
|
+
}
|
|
29801
|
+
return {
|
|
29802
|
+
kept,
|
|
29803
|
+
removedText,
|
|
29804
|
+
removedImages
|
|
29805
|
+
};
|
|
29806
|
+
}
|
|
29807
|
+
/** Replace the page's entire content stream with the given operator string. */
|
|
29808
|
+
function replacePageContent(page, content) {
|
|
29809
|
+
const internals = page;
|
|
29810
|
+
internals.ops = content;
|
|
29811
|
+
internals._originalContentRefs = [];
|
|
29812
|
+
}
|
|
29813
|
+
/**
|
|
29814
|
+
* Permanently remove text and image content within the given regions from a
|
|
29815
|
+
* page (redaction by removal).
|
|
29816
|
+
*
|
|
29817
|
+
* The page's content stream is parsed, replayed through a position-tracking
|
|
29818
|
+
* graphics-state machine, and re-emitted with every text-showing operator and
|
|
29819
|
+
* image placement that falls inside any redaction rect omitted. All other
|
|
29820
|
+
* operators — paths, fills, colors, clips, state changes — are preserved, so
|
|
29821
|
+
* the surrounding page layout is left intact. The filtered content is written
|
|
29822
|
+
* back to the page and takes effect on the next `save()`.
|
|
29823
|
+
*
|
|
29824
|
+
* @param page The page to redact (modified in place).
|
|
29825
|
+
* @param rects Redaction regions in page space (PDF points, y-up).
|
|
29826
|
+
* @returns Counts of removed text runs and image placements.
|
|
29827
|
+
*
|
|
29828
|
+
* @example
|
|
29829
|
+
* ```ts
|
|
29830
|
+
* const result = redactRegions(page, [{ x: 0, y: 90, width: 200, height: 30 }]);
|
|
29831
|
+
* console.log(`Removed ${result.removedText} text runs`);
|
|
29832
|
+
* const bytes = await doc.save();
|
|
29833
|
+
* ```
|
|
29834
|
+
*/
|
|
29835
|
+
function redactRegions(page, rects) {
|
|
29836
|
+
if (rects.length === 0) return {
|
|
29837
|
+
removedText: 0,
|
|
29838
|
+
removedImages: 0
|
|
29839
|
+
};
|
|
29840
|
+
const { kept, removedText, removedImages } = filterOperators(require_compressionAnalysis.parseContentStream(page.getContentStream()), rects);
|
|
29841
|
+
replacePageContent(page, kept.map(serializeOperator).join("\n") + (kept.length > 0 ? "\n" : ""));
|
|
29842
|
+
return {
|
|
29843
|
+
removedText,
|
|
29844
|
+
removedImages
|
|
29845
|
+
};
|
|
29846
|
+
}
|
|
29847
|
+
//#endregion
|
|
29848
|
+
//#region src/render/tiles.ts
|
|
29849
|
+
/**
|
|
29850
|
+
* @module render/tiles
|
|
29851
|
+
*
|
|
29852
|
+
* Tile-based rendering of large pages plus a small LRU cache. A page is
|
|
29853
|
+
* interpreted once into a {@link DisplayList}; each tile rasterizes only its
|
|
29854
|
+
* pixel sub-window via the rasterizer's `region` option, so arbitrarily large
|
|
29855
|
+
* pages render in bounded memory.
|
|
29856
|
+
*
|
|
29857
|
+
* @packageDocumentation
|
|
29858
|
+
*/
|
|
29859
|
+
function scaleOf(opts) {
|
|
29860
|
+
if (typeof opts.scale === "number") return opts.scale;
|
|
29861
|
+
if (typeof opts.dpi === "number") return opts.dpi / 72;
|
|
29862
|
+
return 1;
|
|
29863
|
+
}
|
|
29864
|
+
function gridFromDisplayList(dl, opts) {
|
|
29865
|
+
const scale = scaleOf(opts);
|
|
29866
|
+
const tileSize = opts.tileSize ?? 512;
|
|
29867
|
+
const fullWidth = Math.max(1, Math.ceil(dl.width * scale));
|
|
29868
|
+
const fullHeight = Math.max(1, Math.ceil(dl.height * scale));
|
|
29869
|
+
return {
|
|
29870
|
+
columns: Math.max(1, Math.ceil(fullWidth / tileSize)),
|
|
29871
|
+
rows: Math.max(1, Math.ceil(fullHeight / tileSize)),
|
|
29872
|
+
tileSize,
|
|
29873
|
+
fullWidth,
|
|
29874
|
+
fullHeight,
|
|
29875
|
+
scale
|
|
29876
|
+
};
|
|
29877
|
+
}
|
|
29878
|
+
/** Compute the tile grid for a page. */
|
|
29879
|
+
function computeTileGrid(page, opts = {}) {
|
|
29880
|
+
return gridFromDisplayList(interpretPage(page), opts);
|
|
29881
|
+
}
|
|
29882
|
+
/** Render a single tile `(column, row)` of a page to an RGBA image. */
|
|
29883
|
+
function renderPageTile(page, column, row, opts = {}) {
|
|
29884
|
+
const dl = interpretPage(page);
|
|
29885
|
+
const grid = gridFromDisplayList(dl, opts);
|
|
29886
|
+
const x = column * grid.tileSize;
|
|
29887
|
+
const y = row * grid.tileSize;
|
|
29888
|
+
const width = Math.min(grid.tileSize, grid.fullWidth - x);
|
|
29889
|
+
const height = Math.min(grid.tileSize, grid.fullHeight - y);
|
|
29890
|
+
return rasterize(dl, {
|
|
29891
|
+
scale: grid.scale,
|
|
29892
|
+
background: opts.background,
|
|
29893
|
+
renderText: opts.renderText ?? true,
|
|
29894
|
+
region: {
|
|
29895
|
+
x,
|
|
29896
|
+
y,
|
|
29897
|
+
width: Math.max(1, width),
|
|
29898
|
+
height: Math.max(1, height)
|
|
29899
|
+
}
|
|
29900
|
+
});
|
|
29901
|
+
}
|
|
29902
|
+
/**
|
|
29903
|
+
* A simple least-recently-used cache (Map-backed). Useful for memoizing
|
|
29904
|
+
* interpreted display lists or rasterized tiles across renders.
|
|
29905
|
+
*/
|
|
29906
|
+
var RenderCache = class {
|
|
29907
|
+
maxEntries;
|
|
29908
|
+
map = /* @__PURE__ */ new Map();
|
|
29909
|
+
constructor(maxEntries = 64) {
|
|
29910
|
+
this.maxEntries = maxEntries;
|
|
29911
|
+
}
|
|
29912
|
+
/** Number of cached entries. */
|
|
29913
|
+
get size() {
|
|
29914
|
+
return this.map.size;
|
|
29915
|
+
}
|
|
29916
|
+
/** Look up a value, marking it most-recently-used. */
|
|
29917
|
+
get(key) {
|
|
29918
|
+
const v = this.map.get(key);
|
|
29919
|
+
if (v !== void 0) {
|
|
29920
|
+
this.map.delete(key);
|
|
29921
|
+
this.map.set(key, v);
|
|
29922
|
+
}
|
|
29923
|
+
return v;
|
|
29924
|
+
}
|
|
29925
|
+
/** Whether a key is present (without affecting recency). */
|
|
29926
|
+
has(key) {
|
|
29927
|
+
return this.map.has(key);
|
|
29928
|
+
}
|
|
29929
|
+
/** Insert/replace a value, evicting the LRU entry when over capacity. */
|
|
29930
|
+
set(key, value) {
|
|
29931
|
+
if (this.map.has(key)) this.map.delete(key);
|
|
29932
|
+
this.map.set(key, value);
|
|
29933
|
+
while (this.map.size > this.maxEntries) {
|
|
29934
|
+
const oldest = this.map.keys().next().value;
|
|
29935
|
+
if (oldest === void 0) break;
|
|
29936
|
+
this.map.delete(oldest);
|
|
29937
|
+
}
|
|
29938
|
+
}
|
|
29939
|
+
/** Remove a key. */
|
|
29940
|
+
delete(key) {
|
|
29941
|
+
return this.map.delete(key);
|
|
29942
|
+
}
|
|
29943
|
+
/** Empty the cache. */
|
|
29944
|
+
clear() {
|
|
29945
|
+
this.map.clear();
|
|
29946
|
+
}
|
|
29947
|
+
};
|
|
29948
|
+
//#endregion
|
|
27841
29949
|
//#region src/index.ts
|
|
27842
29950
|
/** Whether initWasm has already completed successfully. */
|
|
27843
29951
|
let wasmInitialized = false;
|
|
@@ -28159,6 +30267,12 @@ Object.defineProperty(exports, "RemovePageFromEmptyDocumentError", {
|
|
|
28159
30267
|
return RemovePageFromEmptyDocumentError;
|
|
28160
30268
|
}
|
|
28161
30269
|
});
|
|
30270
|
+
Object.defineProperty(exports, "RenderCache", {
|
|
30271
|
+
enumerable: true,
|
|
30272
|
+
get: function() {
|
|
30273
|
+
return RenderCache;
|
|
30274
|
+
}
|
|
30275
|
+
});
|
|
28162
30276
|
Object.defineProperty(exports, "RichTextFieldReadError", {
|
|
28163
30277
|
enumerable: true,
|
|
28164
30278
|
get: function() {
|
|
@@ -28261,6 +30375,12 @@ Object.defineProperty(exports, "applyHeaderFooterToPage", {
|
|
|
28261
30375
|
return applyHeaderFooterToPage;
|
|
28262
30376
|
}
|
|
28263
30377
|
});
|
|
30378
|
+
Object.defineProperty(exports, "applyOcr", {
|
|
30379
|
+
enumerable: true,
|
|
30380
|
+
get: function() {
|
|
30381
|
+
return applyOcr;
|
|
30382
|
+
}
|
|
30383
|
+
});
|
|
28264
30384
|
Object.defineProperty(exports, "applyOverflow", {
|
|
28265
30385
|
enumerable: true,
|
|
28266
30386
|
get: function() {
|
|
@@ -28549,6 +30669,18 @@ Object.defineProperty(exports, "code39ToOperators", {
|
|
|
28549
30669
|
return code39ToOperators;
|
|
28550
30670
|
}
|
|
28551
30671
|
});
|
|
30672
|
+
Object.defineProperty(exports, "compareImages", {
|
|
30673
|
+
enumerable: true,
|
|
30674
|
+
get: function() {
|
|
30675
|
+
return compareImages;
|
|
30676
|
+
}
|
|
30677
|
+
});
|
|
30678
|
+
Object.defineProperty(exports, "comparePages", {
|
|
30679
|
+
enumerable: true,
|
|
30680
|
+
get: function() {
|
|
30681
|
+
return comparePages;
|
|
30682
|
+
}
|
|
30683
|
+
});
|
|
28552
30684
|
Object.defineProperty(exports, "computeCode39CheckDigit", {
|
|
28553
30685
|
enumerable: true,
|
|
28554
30686
|
get: function() {
|
|
@@ -28561,6 +30693,12 @@ Object.defineProperty(exports, "computeObjectHash", {
|
|
|
28561
30693
|
return computeObjectHash;
|
|
28562
30694
|
}
|
|
28563
30695
|
});
|
|
30696
|
+
Object.defineProperty(exports, "computeTileGrid", {
|
|
30697
|
+
enumerable: true,
|
|
30698
|
+
get: function() {
|
|
30699
|
+
return computeTileGrid;
|
|
30700
|
+
}
|
|
30701
|
+
});
|
|
28564
30702
|
Object.defineProperty(exports, "convertTiffCmykToRgb", {
|
|
28565
30703
|
enumerable: true,
|
|
28566
30704
|
get: function() {
|
|
@@ -28825,6 +30963,18 @@ Object.defineProperty(exports, "extractEmbeddedRevocationData", {
|
|
|
28825
30963
|
return extractEmbeddedRevocationData;
|
|
28826
30964
|
}
|
|
28827
30965
|
});
|
|
30966
|
+
Object.defineProperty(exports, "extractFonts", {
|
|
30967
|
+
enumerable: true,
|
|
30968
|
+
get: function() {
|
|
30969
|
+
return extractFonts;
|
|
30970
|
+
}
|
|
30971
|
+
});
|
|
30972
|
+
Object.defineProperty(exports, "extractImages", {
|
|
30973
|
+
enumerable: true,
|
|
30974
|
+
get: function() {
|
|
30975
|
+
return extractImages;
|
|
30976
|
+
}
|
|
30977
|
+
});
|
|
28828
30978
|
Object.defineProperty(exports, "extractJpegMetadata", {
|
|
28829
30979
|
enumerable: true,
|
|
28830
30980
|
get: function() {
|
|
@@ -28906,7 +31056,7 @@ Object.defineProperty(exports, "formatDate$1", {
|
|
|
28906
31056
|
Object.defineProperty(exports, "formatNumber", {
|
|
28907
31057
|
enumerable: true,
|
|
28908
31058
|
get: function() {
|
|
28909
|
-
return formatNumber;
|
|
31059
|
+
return formatNumber$1;
|
|
28910
31060
|
}
|
|
28911
31061
|
});
|
|
28912
31062
|
Object.defineProperty(exports, "generateCiiXml", {
|
|
@@ -28993,6 +31143,12 @@ Object.defineProperty(exports, "generateSymbolToUnicodeCmap", {
|
|
|
28993
31143
|
return generateSymbolToUnicodeCmap;
|
|
28994
31144
|
}
|
|
28995
31145
|
});
|
|
31146
|
+
Object.defineProperty(exports, "generateThumbnail", {
|
|
31147
|
+
enumerable: true,
|
|
31148
|
+
get: function() {
|
|
31149
|
+
return generateThumbnail;
|
|
31150
|
+
}
|
|
31151
|
+
});
|
|
28996
31152
|
Object.defineProperty(exports, "generateUnderlineAppearance", {
|
|
28997
31153
|
enumerable: true,
|
|
28998
31154
|
get: function() {
|
|
@@ -29113,6 +31269,18 @@ Object.defineProperty(exports, "injectJpegMetadata", {
|
|
|
29113
31269
|
return injectJpegMetadata;
|
|
29114
31270
|
}
|
|
29115
31271
|
});
|
|
31272
|
+
Object.defineProperty(exports, "interpretContentStream", {
|
|
31273
|
+
enumerable: true,
|
|
31274
|
+
get: function() {
|
|
31275
|
+
return interpretContentStream;
|
|
31276
|
+
}
|
|
31277
|
+
});
|
|
31278
|
+
Object.defineProperty(exports, "interpretPage", {
|
|
31279
|
+
enumerable: true,
|
|
31280
|
+
get: function() {
|
|
31281
|
+
return interpretPage;
|
|
31282
|
+
}
|
|
31283
|
+
});
|
|
29116
31284
|
Object.defineProperty(exports, "isCertificateRevoked", {
|
|
29117
31285
|
enumerable: true,
|
|
29118
31286
|
get: function() {
|
|
@@ -29299,6 +31467,12 @@ Object.defineProperty(exports, "professionalPreset", {
|
|
|
29299
31467
|
return professionalPreset;
|
|
29300
31468
|
}
|
|
29301
31469
|
});
|
|
31470
|
+
Object.defineProperty(exports, "rasterize", {
|
|
31471
|
+
enumerable: true,
|
|
31472
|
+
get: function() {
|
|
31473
|
+
return rasterize;
|
|
31474
|
+
}
|
|
31475
|
+
});
|
|
29302
31476
|
Object.defineProperty(exports, "readBarcode", {
|
|
29303
31477
|
enumerable: true,
|
|
29304
31478
|
get: function() {
|
|
@@ -29359,6 +31533,12 @@ Object.defineProperty(exports, "reconstructParagraphs", {
|
|
|
29359
31533
|
return reconstructParagraphs;
|
|
29360
31534
|
}
|
|
29361
31535
|
});
|
|
31536
|
+
Object.defineProperty(exports, "redactRegions", {
|
|
31537
|
+
enumerable: true,
|
|
31538
|
+
get: function() {
|
|
31539
|
+
return redactRegions;
|
|
31540
|
+
}
|
|
31541
|
+
});
|
|
29362
31542
|
Object.defineProperty(exports, "removeAllBookmarks", {
|
|
29363
31543
|
enumerable: true,
|
|
29364
31544
|
get: function() {
|
|
@@ -29377,6 +31557,30 @@ Object.defineProperty(exports, "renderCodeFrame", {
|
|
|
29377
31557
|
return renderCodeFrame;
|
|
29378
31558
|
}
|
|
29379
31559
|
});
|
|
31560
|
+
Object.defineProperty(exports, "renderDisplayListToCanvas", {
|
|
31561
|
+
enumerable: true,
|
|
31562
|
+
get: function() {
|
|
31563
|
+
return renderDisplayListToCanvas;
|
|
31564
|
+
}
|
|
31565
|
+
});
|
|
31566
|
+
Object.defineProperty(exports, "renderPageTile", {
|
|
31567
|
+
enumerable: true,
|
|
31568
|
+
get: function() {
|
|
31569
|
+
return renderPageTile;
|
|
31570
|
+
}
|
|
31571
|
+
});
|
|
31572
|
+
Object.defineProperty(exports, "renderPageToCanvas", {
|
|
31573
|
+
enumerable: true,
|
|
31574
|
+
get: function() {
|
|
31575
|
+
return renderPageToCanvas;
|
|
31576
|
+
}
|
|
31577
|
+
});
|
|
31578
|
+
Object.defineProperty(exports, "renderPageToImage", {
|
|
31579
|
+
enumerable: true,
|
|
31580
|
+
get: function() {
|
|
31581
|
+
return renderPageToImage;
|
|
31582
|
+
}
|
|
31583
|
+
});
|
|
29380
31584
|
Object.defineProperty(exports, "renderStyledBarcode", {
|
|
29381
31585
|
enumerable: true,
|
|
29382
31586
|
get: function() {
|
|
@@ -29666,4 +31870,4 @@ Object.defineProperty(exports, "wrapText", {
|
|
|
29666
31870
|
}
|
|
29667
31871
|
});
|
|
29668
31872
|
|
|
29669
|
-
//# sourceMappingURL=src-
|
|
31873
|
+
//# sourceMappingURL=src-DQ63jYIR.cjs.map
|