rowpipe 1.0.0 → 1.2.0
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 +327 -41
- package/dist/cli/commands/diff.d.ts +30 -0
- package/dist/cli/commands/diff.d.ts.map +1 -0
- package/dist/cli/commands/diff.js +143 -0
- package/dist/cli/commands/diff.js.map +1 -0
- package/dist/cli/commands/files.d.ts +35 -0
- package/dist/cli/commands/files.d.ts.map +1 -0
- package/dist/cli/commands/files.js +93 -0
- package/dist/cli/commands/files.js.map +1 -0
- package/dist/cli/index.js +89 -5
- package/dist/cli/index.js.map +1 -1
- package/dist/core/errors.d.ts +23 -0
- package/dist/core/errors.d.ts.map +1 -1
- package/dist/core/errors.js +38 -0
- package/dist/core/errors.js.map +1 -1
- package/dist/core/types.d.ts +4 -0
- package/dist/core/types.d.ts.map +1 -1
- package/dist/diff/comparator.d.ts +17 -0
- package/dist/diff/comparator.d.ts.map +1 -0
- package/dist/diff/comparator.js +158 -0
- package/dist/diff/comparator.js.map +1 -0
- package/dist/diff/engine.d.ts +10 -0
- package/dist/diff/engine.d.ts.map +1 -0
- package/dist/diff/engine.js +333 -0
- package/dist/diff/engine.js.map +1 -0
- package/dist/diff/hash.d.ts +6 -0
- package/dist/diff/hash.d.ts.map +1 -0
- package/dist/diff/hash.js +50 -0
- package/dist/diff/hash.js.map +1 -0
- package/dist/diff/key.d.ts +21 -0
- package/dist/diff/key.d.ts.map +1 -0
- package/dist/diff/key.js +107 -0
- package/dist/diff/key.js.map +1 -0
- package/dist/diff/reporter.d.ts +18 -0
- package/dist/diff/reporter.d.ts.map +1 -0
- package/dist/diff/reporter.js +133 -0
- package/dist/diff/reporter.js.map +1 -0
- package/dist/diff/schema.d.ts +7 -0
- package/dist/diff/schema.d.ts.map +1 -0
- package/dist/diff/schema.js +65 -0
- package/dist/diff/schema.js.map +1 -0
- package/dist/diff/storage/disk-index.d.ts +28 -0
- package/dist/diff/storage/disk-index.d.ts.map +1 -0
- package/dist/diff/storage/disk-index.js +105 -0
- package/dist/diff/storage/disk-index.js.map +1 -0
- package/dist/diff/storage/memory-index.d.ts +23 -0
- package/dist/diff/storage/memory-index.d.ts.map +1 -0
- package/dist/diff/storage/memory-index.js +69 -0
- package/dist/diff/storage/memory-index.js.map +1 -0
- package/dist/diff/storage/spillable-index.d.ts +30 -0
- package/dist/diff/storage/spillable-index.d.ts.map +1 -0
- package/dist/diff/storage/spillable-index.js +108 -0
- package/dist/diff/storage/spillable-index.js.map +1 -0
- package/dist/diff/types.d.ts +127 -0
- package/dist/diff/types.d.ts.map +1 -0
- package/dist/diff/types.js +2 -0
- package/dist/diff/types.js.map +1 -0
- package/dist/files/glob.d.ts +21 -0
- package/dist/files/glob.d.ts.map +1 -0
- package/dist/files/glob.js +132 -0
- package/dist/files/glob.js.map +1 -0
- package/dist/files/hash.d.ts +10 -0
- package/dist/files/hash.d.ts.map +1 -0
- package/dist/files/hash.js +67 -0
- package/dist/files/hash.js.map +1 -0
- package/dist/files/index.d.ts +6 -0
- package/dist/files/index.d.ts.map +1 -0
- package/dist/files/index.js +6 -0
- package/dist/files/index.js.map +1 -0
- package/dist/files/mime.d.ts +9 -0
- package/dist/files/mime.d.ts.map +1 -0
- package/dist/files/mime.js +128 -0
- package/dist/files/mime.js.map +1 -0
- package/dist/files/reader.d.ts +19 -0
- package/dist/files/reader.d.ts.map +1 -0
- package/dist/files/reader.js +344 -0
- package/dist/files/reader.js.map +1 -0
- package/dist/files/types.d.ts +47 -0
- package/dist/files/types.d.ts.map +1 -0
- package/dist/files/types.js +2 -0
- package/dist/files/types.js.map +1 -0
- package/dist/index.d.ts +17 -2
- package/dist/index.d.ts.map +1 -1
- package/dist/index.js +19 -2
- package/dist/index.js.map +1 -1
- package/dist/readers/csv.d.ts.map +1 -1
- package/dist/readers/csv.js +59 -27
- package/dist/readers/csv.js.map +1 -1
- package/dist/readers/index.d.ts +2 -0
- package/dist/readers/index.d.ts.map +1 -1
- package/dist/readers/index.js +57 -3
- package/dist/readers/index.js.map +1 -1
- package/dist/readers/json.d.ts.map +1 -1
- package/dist/readers/json.js +6 -8
- package/dist/readers/json.js.map +1 -1
- package/dist/readers/jsonl.d.ts.map +1 -1
- package/dist/readers/jsonl.js +5 -7
- package/dist/readers/jsonl.js.map +1 -1
- package/dist/readers/parquet.d.ts +17 -0
- package/dist/readers/parquet.d.ts.map +1 -0
- package/dist/readers/parquet.js +146 -0
- package/dist/readers/parquet.js.map +1 -0
- package/dist/transforms/expression.d.ts.map +1 -1
- package/dist/transforms/expression.js +73 -7
- package/dist/transforms/expression.js.map +1 -1
- package/dist/utils/compression.d.ts +30 -6
- package/dist/utils/compression.d.ts.map +1 -1
- package/dist/utils/compression.js +95 -18
- package/dist/utils/compression.js.map +1 -1
- package/dist/writers/csv.d.ts.map +1 -1
- package/dist/writers/csv.js +29 -17
- package/dist/writers/csv.js.map +1 -1
- package/dist/writers/index.d.ts +2 -0
- package/dist/writers/index.d.ts.map +1 -1
- package/dist/writers/index.js +49 -4
- package/dist/writers/index.js.map +1 -1
- package/dist/writers/json.d.ts +1 -1
- package/dist/writers/json.d.ts.map +1 -1
- package/dist/writers/json.js +9 -11
- package/dist/writers/json.js.map +1 -1
- package/dist/writers/jsonl.d.ts +1 -1
- package/dist/writers/jsonl.d.ts.map +1 -1
- package/dist/writers/jsonl.js +9 -11
- package/dist/writers/jsonl.js.map +1 -1
- package/dist/writers/markdown.d.ts +21 -0
- package/dist/writers/markdown.d.ts.map +1 -0
- package/dist/writers/markdown.js +82 -0
- package/dist/writers/markdown.js.map +1 -0
- package/dist/writers/parquet.d.ts +19 -0
- package/dist/writers/parquet.d.ts.map +1 -0
- package/dist/writers/parquet.js +129 -0
- package/dist/writers/parquet.js.map +1 -0
- package/package.json +16 -3
- package/skills/rowpipe/SKILL.md +263 -0
- package/benchmarks/memory-bench.ts +0 -109
- package/src/analytics/reduce.ts +0 -388
- package/src/analytics/schema-inference.ts +0 -217
- package/src/analytics/semantic-types.ts +0 -36
- package/src/analytics/stats.ts +0 -383
- package/src/analytics/validator.ts +0 -134
- package/src/cli/commands/cast.ts +0 -60
- package/src/cli/commands/convert.ts +0 -128
- package/src/cli/commands/filter.ts +0 -58
- package/src/cli/commands/inspect.ts +0 -193
- package/src/cli/commands/map.ts +0 -66
- package/src/cli/commands/reduce.ts +0 -74
- package/src/cli/commands/rename.ts +0 -59
- package/src/cli/commands/sample.ts +0 -61
- package/src/cli/commands/schema.ts +0 -86
- package/src/cli/commands/select.ts +0 -59
- package/src/cli/commands/stats.ts +0 -100
- package/src/cli/commands/validate.ts +0 -123
- package/src/cli/index.ts +0 -285
- package/src/core/batch.ts +0 -81
- package/src/core/errors.ts +0 -81
- package/src/core/pipeline.ts +0 -150
- package/src/core/types.ts +0 -110
- package/src/index.ts +0 -52
- package/src/readers/csv.ts +0 -302
- package/src/readers/index.ts +0 -88
- package/src/readers/json.ts +0 -256
- package/src/readers/jsonl.ts +0 -126
- package/src/readers/xlsx.ts +0 -182
- package/src/transforms/cast.ts +0 -212
- package/src/transforms/expression.ts +0 -1114
- package/src/transforms/filter.ts +0 -38
- package/src/transforms/map.ts +0 -99
- package/src/transforms/rename.ts +0 -47
- package/src/transforms/sample.ts +0 -57
- package/src/transforms/select.ts +0 -64
- package/src/utils/compression.ts +0 -43
- package/src/utils/formatting.ts +0 -75
- package/src/utils/progress.ts +0 -37
- package/src/writers/csv.ts +0 -106
- package/src/writers/index.ts +0 -87
- package/src/writers/json.ts +0 -52
- package/src/writers/jsonl.ts +0 -49
- package/src/writers/xlsx.ts +0 -87
- package/tests/analytics.test.ts +0 -119
- package/tests/cli-integration.test.ts +0 -187
- package/tests/map-reduce.test.ts +0 -223
- package/tests/readers-writers.test.ts +0 -189
- package/tests/transforms.test.ts +0 -199
- package/tsconfig.json +0 -21
- package/vitest.config.ts +0 -9
package/dist/readers/csv.js
CHANGED
|
@@ -1,4 +1,4 @@
|
|
|
1
|
-
import {
|
|
1
|
+
import { openDecompressedReadStream } from "../utils/compression.js";
|
|
2
2
|
import { ParseError } from "../core/errors.js";
|
|
3
3
|
/**
|
|
4
4
|
* Streaming CSV Reader supporting RFC 4180 quotes, multiline cells,
|
|
@@ -14,11 +14,9 @@ export class CSVReader {
|
|
|
14
14
|
this.options.filePath = input;
|
|
15
15
|
}
|
|
16
16
|
}
|
|
17
|
-
getInputStream() {
|
|
18
|
-
|
|
19
|
-
|
|
20
|
-
}
|
|
21
|
-
return this.input;
|
|
17
|
+
getInputStream(opts) {
|
|
18
|
+
const gzip = opts?.gzip ?? this.options.gzip;
|
|
19
|
+
return openDecompressedReadStream(this.input, { gzip });
|
|
22
20
|
}
|
|
23
21
|
/**
|
|
24
22
|
* Auto-detects delimiter by scanning candidate delimiters in a sample string.
|
|
@@ -55,7 +53,7 @@ export class CSVReader {
|
|
|
55
53
|
const batchSize = Math.max(1, mergedOptions.batchSize ?? 1000);
|
|
56
54
|
const hasHeader = mergedOptions.header !== false;
|
|
57
55
|
let delimiter = mergedOptions.delimiter;
|
|
58
|
-
const stream = this.getInputStream();
|
|
56
|
+
const stream = this.getInputStream(mergedOptions);
|
|
59
57
|
let buffer = "";
|
|
60
58
|
let headers = null;
|
|
61
59
|
let isFirstChunk = true;
|
|
@@ -88,36 +86,51 @@ export class CSVReader {
|
|
|
88
86
|
}
|
|
89
87
|
const activeDelimiter = delimiter ?? ",";
|
|
90
88
|
const delimLen = activeDelimiter.length;
|
|
89
|
+
const isSingleCharDelim = delimLen === 1;
|
|
90
|
+
const delimCode = activeDelimiter.charCodeAt(0);
|
|
91
|
+
let fieldStart = cursor;
|
|
91
92
|
while (cursor < buffer.length) {
|
|
92
|
-
const char = buffer[cursor];
|
|
93
93
|
if (inQuotes) {
|
|
94
|
-
|
|
95
|
-
|
|
96
|
-
|
|
97
|
-
cursor
|
|
98
|
-
|
|
99
|
-
|
|
94
|
+
const charCode = buffer.charCodeAt(cursor);
|
|
95
|
+
if (charCode === 34 /* " */) {
|
|
96
|
+
if (cursor + 1 < buffer.length) {
|
|
97
|
+
if (buffer.charCodeAt(cursor + 1) === 34) {
|
|
98
|
+
// Escaped double quote ("")
|
|
99
|
+
currentField += buffer.slice(fieldStart, cursor) + '"';
|
|
100
|
+
cursor += 2;
|
|
101
|
+
byteOffset += 2;
|
|
102
|
+
fieldStart = cursor;
|
|
103
|
+
continue;
|
|
104
|
+
}
|
|
105
|
+
else {
|
|
106
|
+
// Closing quote
|
|
107
|
+
currentField += buffer.slice(fieldStart, cursor);
|
|
108
|
+
inQuotes = false;
|
|
109
|
+
cursor++;
|
|
110
|
+
byteOffset++;
|
|
111
|
+
fieldStart = cursor;
|
|
112
|
+
continue;
|
|
113
|
+
}
|
|
100
114
|
}
|
|
101
115
|
else {
|
|
102
|
-
|
|
103
|
-
|
|
104
|
-
byteOffset++;
|
|
105
|
-
continue;
|
|
116
|
+
// Reached end of current buffer inside quotes; wait for next chunk
|
|
117
|
+
break;
|
|
106
118
|
}
|
|
107
119
|
}
|
|
108
120
|
else {
|
|
109
|
-
currentField += char;
|
|
110
121
|
cursor++;
|
|
111
122
|
byteOffset++;
|
|
112
123
|
continue;
|
|
113
124
|
}
|
|
114
125
|
}
|
|
115
126
|
else {
|
|
116
|
-
|
|
117
|
-
|
|
127
|
+
const charCode = buffer.charCodeAt(cursor);
|
|
128
|
+
if (charCode === 34 /* " */) {
|
|
129
|
+
if (cursor === fieldStart && currentField.length === 0) {
|
|
118
130
|
inQuotes = true;
|
|
119
131
|
cursor++;
|
|
120
132
|
byteOffset++;
|
|
133
|
+
fieldStart = cursor;
|
|
121
134
|
continue;
|
|
122
135
|
}
|
|
123
136
|
else {
|
|
@@ -130,18 +143,34 @@ export class CSVReader {
|
|
|
130
143
|
}
|
|
131
144
|
}
|
|
132
145
|
// Check for delimiter
|
|
133
|
-
|
|
134
|
-
|
|
146
|
+
const isDelim = isSingleCharDelim
|
|
147
|
+
? charCode === delimCode
|
|
148
|
+
: (buffer.length - cursor >= delimLen && buffer.startsWith(activeDelimiter, cursor));
|
|
149
|
+
if (isDelim) {
|
|
150
|
+
const fieldPart = buffer.slice(fieldStart, cursor);
|
|
151
|
+
const val = currentField.length > 0 ? currentField + fieldPart : fieldPart;
|
|
152
|
+
currentRowFields.push(val);
|
|
135
153
|
currentField = "";
|
|
136
154
|
cursor += delimLen;
|
|
137
155
|
byteOffset += delimLen;
|
|
156
|
+
fieldStart = cursor;
|
|
138
157
|
continue;
|
|
139
158
|
}
|
|
159
|
+
// If remaining buffer is shorter than multi-char delimiter, wait for next chunk
|
|
160
|
+
if (!isSingleCharDelim && buffer.length - cursor < delimLen) {
|
|
161
|
+
break;
|
|
162
|
+
}
|
|
140
163
|
// Check for CRLF / LF newline
|
|
141
|
-
if (
|
|
142
|
-
|
|
164
|
+
if (charCode === 13 /* \r */ || charCode === 10 /* \n */) {
|
|
165
|
+
if (charCode === 13 && cursor + 1 >= buffer.length) {
|
|
166
|
+
// Wait for next chunk to check for \n
|
|
167
|
+
break;
|
|
168
|
+
}
|
|
169
|
+
const fieldPart = buffer.slice(fieldStart, cursor);
|
|
170
|
+
const val = currentField.length > 0 ? currentField + fieldPart : fieldPart;
|
|
171
|
+
currentRowFields.push(val);
|
|
143
172
|
currentField = "";
|
|
144
|
-
if (
|
|
173
|
+
if (charCode === 13 && cursor + 1 < buffer.length && buffer.charCodeAt(cursor + 1) === 10) {
|
|
145
174
|
cursor += 2;
|
|
146
175
|
byteOffset += 2;
|
|
147
176
|
}
|
|
@@ -149,6 +178,7 @@ export class CSVReader {
|
|
|
149
178
|
cursor++;
|
|
150
179
|
byteOffset++;
|
|
151
180
|
}
|
|
181
|
+
fieldStart = cursor;
|
|
152
182
|
// Skip entirely empty row if it's trailing whitespace
|
|
153
183
|
if (currentRowFields.length === 1 &&
|
|
154
184
|
currentRowFields[0]?.trim() === "" &&
|
|
@@ -184,12 +214,14 @@ export class CSVReader {
|
|
|
184
214
|
}
|
|
185
215
|
continue;
|
|
186
216
|
}
|
|
187
|
-
currentField += char;
|
|
188
217
|
cursor++;
|
|
189
218
|
byteOffset++;
|
|
190
219
|
}
|
|
191
220
|
}
|
|
192
221
|
// Compact processed buffer to keep memory bounded
|
|
222
|
+
if (fieldStart < cursor) {
|
|
223
|
+
currentField += buffer.slice(fieldStart, cursor);
|
|
224
|
+
}
|
|
193
225
|
buffer = buffer.slice(cursor);
|
|
194
226
|
cursor = 0;
|
|
195
227
|
}
|
package/dist/readers/csv.js.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"csv.js","sourceRoot":"","sources":["../../src/readers/csv.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,
|
|
1
|
+
{"version":3,"file":"csv.js","sourceRoot":"","sources":["../../src/readers/csv.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,0BAA0B,EAAE,MAAM,yBAAyB,CAAC;AAErE,OAAO,EAAE,UAAU,EAAE,MAAM,mBAAmB,CAAC;AAgB/C;;;GAGG;AACH,MAAM,OAAO,SAAS;IACZ,KAAK,CAAoB;IACzB,OAAO,CAAmB;IAElC,YAAY,KAAwB,EAAE,UAA4B,EAAE;QAClE,IAAI,CAAC,KAAK,GAAG,KAAK,CAAC;QACnB,IAAI,CAAC,OAAO,GAAG,EAAE,GAAG,OAAO,EAAE,CAAC;QAC9B,IAAI,OAAO,KAAK,KAAK,QAAQ,EAAE,CAAC;YAC9B,IAAI,CAAC,OAAO,CAAC,QAAQ,GAAG,KAAK,CAAC;QAChC,CAAC;IACH,CAAC;IAEO,cAAc,CAAC,IAAoB;QACzC,MAAM,IAAI,GAAG,IAAI,EAAE,IAAI,IAAI,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC;QAC7C,OAAO,0BAA0B,CAAC,IAAI,CAAC,KAAK,EAAE,EAAE,IAAI,EAAE,CAAC,CAAC;IAC1D,CAAC;IAED;;OAEG;IACI,MAAM,CAAC,eAAe,CAAC,MAAc;QAC1C,MAAM,UAAU,GAAG,CAAC,GAAG,EAAE,IAAI,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC;QACzC,MAAM,SAAS,GAAG,MAAM,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;QAEjD,IAAI,aAAa,GAAG,GAAG,CAAC;QACxB,IAAI,QAAQ,GAAG,CAAC,CAAC,CAAC;QAElB,KAAK,MAAM,IAAI,IAAI,UAAU,EAAE,CAAC;YAC9B,IAAI,KAAK,GAAG,CAAC,CAAC;YACd,IAAI,QAAQ,GAAG,KAAK,CAAC;YACrB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,SAAS,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;gBAC1C,MAAM,EAAE,GAAG,SAAS,CAAC,CAAC,CAAC,CAAC;gBACxB,IAAI,EAAE,KAAK,GAAG,EAAE,CAAC;oBACf,QAAQ,GAAG,CAAC,QAAQ,CAAC;gBACvB,CAAC;qBAAM,IAAI,EAAE,KAAK,IAAI,IAAI,CAAC,QAAQ,EAAE,CAAC;oBACpC,KAAK,EAAE,CAAC;gBACV,CAAC;YACH,CAAC;YACD,IAAI,KAAK,GAAG,QAAQ,EAAE,CAAC;gBACrB,QAAQ,GAAG,KAAK,CAAC;gBACjB,aAAa,GAAG,IAAI,CAAC;YACvB,CAAC;QACH,CAAC;QAED,OAAO,QAAQ,GAAG,CAAC,CAAC,CAAC,CAAC,aAAa,CAAC,CAAC,CAAC,GAAG,CAAC;IAC5C,CAAC;IAED,KAAK,CAAC,CAAC,IAAI,CAAC,OAAuB;QACjC,MAAM,aAAa,GAAqB;YACtC,GAAG,IAAI,CAAC,OAAO;YACf,GAAG,OAAO;SACX,CAAC;QACF,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,aAAa,CAAC,SAAS,IAAI,IAAI,CAAC,CAAC;QAC/D,MAAM,SAAS,GAAG,aAAa,CAAC,MAAM,KAAK,KAAK,CAAC;QACjD,IAAI,SAAS,GAAG,aAAa,CAAC,SAAS,CAAC;QAExC,MAAM,MAAM,GAAG,IAAI,CAAC,cAAc,CAAC,aAAa,CAAC,CAAC;QAElD,IAAI,MAAM,GAAG,EAAE,CAAC;QAChB,IAAI,OAAO,GAAoB,IAAI,CAAC;QACpC,IAAI,YAAY,GAAG,IAAI,CAAC;QACxB,IAAI,kBAAkB,GAAU,EAAE,CAAC;QACnC,IAAI,eAAe,GAAG,CAAC,CAAC;QACxB,IAAI,SAAS,GAAG,CAAC,CAAC;QAClB,IAAI,UAAU,GAAG,CAAC,CAAC;QACnB,IAAI,YAAY,GAAG,EAAE,CAAC;QACtB,IAAI,gBAAgB,GAAa,EAAE,CAAC;QACpC,IAAI,QAAQ,GAAG,KAAK,CAAC;QACrB,IAAI,MAAM,GAAG,CAAC,CAAC;QAEf,IAAI,KAAK,EAAE,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;YACjC,MAAM,QAAQ,GACZ,OAAO,KAAK,KAAK,QAAQ,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAE,KAAgB,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC;YAEzE,IAAI,YAAY,EAAE,CAAC;gBACjB,YAAY,GAAG,KAAK,CAAC;gBACrB,8BAA8B;gBAC9B,IAAI,QAAQ,CAAC,UAAU,CAAC,CAAC,CAAC,KAAK,MAAM,EAAE,CAAC;oBACtC,MAAM,GAAG,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;gBAC7B,CAAC;qBAAM,CAAC;oBACN,MAAM,GAAG,QAAQ,CAAC;gBACpB,CAAC;gBAED,yCAAyC;gBACzC,IAAI,CAAC,SAAS,EAAE,CAAC;oBACf,SAAS,GAAG,SAAS,CAAC,eAAe,CAAC,MAAM,CAAC,CAAC;gBAChD,CAAC;YACH,CAAC;iBAAM,CAAC;gBACN,MAAM,IAAI,QAAQ,CAAC;YACrB,CAAC;YAED,MAAM,eAAe,GAAG,SAAS,IAAI,GAAG,CAAC;YACzC,MAAM,QAAQ,GAAG,eAAe,CAAC,MAAM,CAAC;YACxC,MAAM,iBAAiB,GAAG,QAAQ,KAAK,CAAC,CAAC;YACzC,MAAM,SAAS,GAAG,eAAe,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC;YAEhD,IAAI,UAAU,GAAG,MAAM,CAAC;YAExB,OAAO,MAAM,GAAG,MAAM,CAAC,MAAM,EAAE,CAAC;gBAC9B,IAAI,QAAQ,EAAE,CAAC;oBACb,MAAM,QAAQ,GAAG,MAAM,CAAC,UAAU,CAAC,MAAM,CAAC,CAAC;oBAC3C,IAAI,QAAQ,KAAK,EAAE,CAAC,OAAO,EAAE,CAAC;wBAC5B,IAAI,MAAM,GAAG,CAAC,GAAG,MAAM,CAAC,MAAM,EAAE,CAAC;4BAC/B,IAAI,MAAM,CAAC,UAAU,CAAC,MAAM,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC;gCACzC,4BAA4B;gCAC5B,YAAY,IAAI,MAAM,CAAC,KAAK,CAAC,UAAU,EAAE,MAAM,CAAC,GAAG,GAAG,CAAC;gCACvD,MAAM,IAAI,CAAC,CAAC;gCACZ,UAAU,IAAI,CAAC,CAAC;gCAChB,UAAU,GAAG,MAAM,CAAC;gCACpB,SAAS;4BACX,CAAC;iCAAM,CAAC;gCACN,gBAAgB;gCAChB,YAAY,IAAI,MAAM,CAAC,KAAK,CAAC,UAAU,EAAE,MAAM,CAAC,CAAC;gCACjD,QAAQ,GAAG,KAAK,CAAC;gCACjB,MAAM,EAAE,CAAC;gCACT,UAAU,EAAE,CAAC;gCACb,UAAU,GAAG,MAAM,CAAC;gCACpB,SAAS;4BACX,CAAC;wBACH,CAAC;6BAAM,CAAC;4BACN,mEAAmE;4BACnE,MAAM;wBACR,CAAC;oBACH,CAAC;yBAAM,CAAC;wBACN,MAAM,EAAE,CAAC;wBACT,UAAU,EAAE,CAAC;wBACb,SAAS;oBACX,CAAC;gBACH,CAAC;qBAAM,CAAC;oBACN,MAAM,QAAQ,GAAG,MAAM,CAAC,UAAU,CAAC,MAAM,CAAC,CAAC;oBAE3C,IAAI,QAAQ,KAAK,EAAE,CAAC,OAAO,EAAE,CAAC;wBAC5B,IAAI,MAAM,KAAK,UAAU,IAAI,YAAY,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;4BACvD,QAAQ,GAAG,IAAI,CAAC;4BAChB,MAAM,EAAE,CAAC;4BACT,UAAU,EAAE,CAAC;4BACb,UAAU,GAAG,MAAM,CAAC;4BACpB,SAAS;wBACX,CAAC;6BAAM,CAAC;4BACN,MAAM,IAAI,UAAU,CAClB,yDAAyD,UAAU,EAAE,EACrE;gCACE,IAAI,EAAE,aAAa,CAAC,QAAQ;gCAC5B,GAAG,EAAE,SAAS,GAAG,CAAC;gCAClB,MAAM,EAAE,gBAAgB,CAAC,MAAM,GAAG,CAAC;gCACnC,UAAU;6BACX,CACF,CAAC;wBACJ,CAAC;oBACH,CAAC;oBAED,sBAAsB;oBACtB,MAAM,OAAO,GAAG,iBAAiB;wBAC/B,CAAC,CAAC,QAAQ,KAAK,SAAS;wBACxB,CAAC,CAAC,CAAC,MAAM,CAAC,MAAM,GAAG,MAAM,IAAI,QAAQ,IAAI,MAAM,CAAC,UAAU,CAAC,eAAe,EAAE,MAAM,CAAC,CAAC,CAAC;oBAEvF,IAAI,OAAO,EAAE,CAAC;wBACZ,MAAM,SAAS,GAAG,MAAM,CAAC,KAAK,CAAC,UAAU,EAAE,MAAM,CAAC,CAAC;wBACnD,MAAM,GAAG,GAAG,YAAY,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,YAAY,GAAG,SAAS,CAAC,CAAC,CAAC,SAAS,CAAC;wBAC3E,gBAAgB,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;wBAC3B,YAAY,GAAG,EAAE,CAAC;wBAClB,MAAM,IAAI,QAAQ,CAAC;wBACnB,UAAU,IAAI,QAAQ,CAAC;wBACvB,UAAU,GAAG,MAAM,CAAC;wBACpB,SAAS;oBACX,CAAC;oBAED,gFAAgF;oBAChF,IAAI,CAAC,iBAAiB,IAAI,MAAM,CAAC,MAAM,GAAG,MAAM,GAAG,QAAQ,EAAE,CAAC;wBAC5D,MAAM;oBACR,CAAC;oBAED,8BAA8B;oBAC9B,IAAI,QAAQ,KAAK,EAAE,CAAC,QAAQ,IAAI,QAAQ,KAAK,EAAE,CAAC,QAAQ,EAAE,CAAC;wBACzD,IAAI,QAAQ,KAAK,EAAE,IAAI,MAAM,GAAG,CAAC,IAAI,MAAM,CAAC,MAAM,EAAE,CAAC;4BACnD,sCAAsC;4BACtC,MAAM;wBACR,CAAC;wBAED,MAAM,SAAS,GAAG,MAAM,CAAC,KAAK,CAAC,UAAU,EAAE,MAAM,CAAC,CAAC;wBACnD,MAAM,GAAG,GAAG,YAAY,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,YAAY,GAAG,SAAS,CAAC,CAAC,CAAC,SAAS,CAAC;wBAC3E,gBAAgB,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;wBAC3B,YAAY,GAAG,EAAE,CAAC;wBAElB,IAAI,QAAQ,KAAK,EAAE,IAAI,MAAM,GAAG,CAAC,GAAG,MAAM,CAAC,MAAM,IAAI,MAAM,CAAC,UAAU,CAAC,MAAM,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC;4BAC1F,MAAM,IAAI,CAAC,CAAC;4BACZ,UAAU,IAAI,CAAC,CAAC;wBAClB,CAAC;6BAAM,CAAC;4BACN,MAAM,EAAE,CAAC;4BACT,UAAU,EAAE,CAAC;wBACf,CAAC;wBACD,UAAU,GAAG,MAAM,CAAC;wBAEpB,sDAAsD;wBACtD,IACE,gBAAgB,CAAC,MAAM,KAAK,CAAC;4BAC7B,gBAAgB,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE;4BAClC,MAAM,IAAI,MAAM,CAAC,MAAM,EACvB,CAAC;4BACD,gBAAgB,GAAG,EAAE,CAAC;4BACtB,SAAS;wBACX,CAAC;wBAED,IAAI,SAAS,IAAI,OAAO,KAAK,IAAI,EAAE,CAAC;4BAClC,OAAO,GAAG,gBAAgB,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CACtC,CAAC,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,OAAO,CAAC,GAAG,CAAC,EAAE,CAChD,CAAC;4BACF,gBAAgB,GAAG,EAAE,CAAC;4BACtB,SAAS;wBACX,CAAC;wBAED,IAAI,OAAO,KAAK,IAAI,EAAE,CAAC;4BACrB,OAAO,GAAG,gBAAgB,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,OAAO,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;wBAC3D,CAAC;wBAED,uBAAuB;wBACvB,MAAM,MAAM,GAAQ,EAAE,CAAC;wBACvB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,OAAO,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;4BACxC,MAAM,GAAG,GAAG,OAAO,CAAC,CAAC,CAAC,IAAI,OAAO,CAAC,GAAG,CAAC,EAAE,CAAC;4BACzC,MAAM,GAAG,GAAG,gBAAgB,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;4BACtC,MAAM,CAAC,GAAG,CAAC,GAAG,GAAG,CAAC;wBACpB,CAAC;wBAED,SAAS,EAAE,CAAC;wBACZ,kBAAkB,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;wBAChC,gBAAgB,GAAG,EAAE,CAAC;wBAEtB,IAAI,kBAAkB,CAAC,MAAM,IAAI,SAAS,EAAE,CAAC;4BAC3C,MAAM;gCACJ,IAAI,EAAE,kBAAkB;gCACxB,MAAM,EAAE,eAAe;6BACxB,CAAC;4BACF,eAAe,IAAI,kBAAkB,CAAC,MAAM,CAAC;4BAC7C,kBAAkB,GAAG,EAAE,CAAC;wBAC1B,CAAC;wBACD,SAAS;oBACX,CAAC;oBAED,MAAM,EAAE,CAAC;oBACT,UAAU,EAAE,CAAC;gBACf,CAAC;YACH,CAAC;YAED,kDAAkD;YAClD,IAAI,UAAU,GAAG,MAAM,EAAE,CAAC;gBACxB,YAAY,IAAI,MAAM,CAAC,KAAK,CAAC,UAAU,EAAE,MAAM,CAAC,CAAC;YACnD,CAAC;YACD,MAAM,GAAG,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC;YAC9B,MAAM,GAAG,CAAC,CAAC;QACb,CAAC;QAED,2BAA2B;QAC3B,IAAI,QAAQ,EAAE,CAAC;YACb,MAAM,IAAI,UAAU,CAAC,qCAAqC,EAAE;gBAC1D,IAAI,EAAE,aAAa,CAAC,QAAQ;gBAC5B,GAAG,EAAE,SAAS,GAAG,CAAC;gBAClB,UAAU;aACX,CAAC,CAAC;QACL,CAAC;QAED,IAAI,YAAY,CAAC,MAAM,GAAG,CAAC,IAAI,gBAAgB,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YAC3D,gBAAgB,CAAC,IAAI,CAAC,YAAY,CAAC,CAAC;YAEpC,IAAI,SAAS,IAAI,OAAO,KAAK,IAAI,EAAE,CAAC;gBAClC,OAAO,GAAG,gBAAgB,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CACtC,CAAC,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,OAAO,CAAC,GAAG,CAAC,EAAE,CAChD,CAAC;YACJ,CAAC;iBAAM,CAAC;gBACN,IAAI,OAAO,KAAK,IAAI,EAAE,CAAC;oBACrB,OAAO,GAAG,gBAAgB,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,OAAO,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;gBAC3D,CAAC;gBAED,MAAM,MAAM,GAAQ,EAAE,CAAC;gBACvB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,OAAO,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;oBACxC,MAAM,GAAG,GAAG,OAAO,CAAC,CAAC,CAAC,IAAI,OAAO,CAAC,GAAG,CAAC,EAAE,CAAC;oBACzC,MAAM,GAAG,GAAG,gBAAgB,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;oBACtC,MAAM,CAAC,GAAG,CAAC,GAAG,GAAG,CAAC;gBACpB,CAAC;gBAED,SAAS,EAAE,CAAC;gBACZ,kBAAkB,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;YAClC,CAAC;QACH,CAAC;QAED,IAAI,kBAAkB,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YAClC,MAAM;gBACJ,IAAI,EAAE,kBAAkB;gBACxB,MAAM,EAAE,eAAe;aACxB,CAAC;QACJ,CAAC;IACH,CAAC;IAED,KAAK,CAAC,OAAO,CAAC,OAAuB;QACnC,IAAI,QAAQ,GAAG,CAAC,CAAC;QACjB,IAAI,WAAW,GAAa,EAAE,CAAC;QAE/B,IAAI,KAAK,EAAE,MAAM,KAAK,IAAI,IAAI,CAAC,IAAI,CAAC,EAAE,GAAG,OAAO,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,EAAE,CAAC;YACrE,IAAI,WAAW,CAAC,MAAM,KAAK,CAAC,IAAI,KAAK,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,IAAI,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC;gBACvE,WAAW,GAAG,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;YAC3C,CAAC;YACD,QAAQ,IAAI,KAAK,CAAC,IAAI,CAAC,MAAM,CAAC;QAChC,CAAC;QAED,OAAO;YACL,MAAM,EAAE,KAAK;YACb,QAAQ;YACR,WAAW,EAAE,WAAW,CAAC,MAAM;YAC/B,OAAO,EAAE,WAAW,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC;gBAClC,IAAI;gBACJ,IAAI,EAAE,QAAQ;gBACd,cAAc,EAAE,CAAC;aAClB,CAAC,CAAC;SACJ,CAAC;IACJ,CAAC;CACF"}
|
package/dist/readers/index.d.ts
CHANGED
|
@@ -1,8 +1,10 @@
|
|
|
1
1
|
import type { Readable } from "node:stream";
|
|
2
2
|
import type { ReaderOptions, TabularReader } from "../core/types.js";
|
|
3
|
+
export * from "../files/reader.js";
|
|
3
4
|
export * from "./csv.js";
|
|
4
5
|
export * from "./json.js";
|
|
5
6
|
export * from "./jsonl.js";
|
|
7
|
+
export * from "./parquet.js";
|
|
6
8
|
export * from "./xlsx.js";
|
|
7
9
|
/**
|
|
8
10
|
* Infers format name from file path or extension.
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../src/readers/index.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,aAAa,CAAC;AAE5C,OAAO,KAAK,EAAiB,aAAa,EAAE,aAAa,EAAE,MAAM,kBAAkB,CAAC;
|
|
1
|
+
{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../src/readers/index.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,aAAa,CAAC;AAE5C,OAAO,KAAK,EAAiB,aAAa,EAAE,aAAa,EAAE,MAAM,kBAAkB,CAAC;AASpF,cAAc,oBAAoB,CAAC;AACnC,cAAc,UAAU,CAAC;AACzB,cAAc,WAAW,CAAC;AAC1B,cAAc,YAAY,CAAC;AAC3B,cAAc,cAAc,CAAC;AAC7B,cAAc,WAAW,CAAC;AAyF1B;;GAEG;AACH,wBAAgB,mBAAmB,CAAC,QAAQ,EAAE,MAAM,GAAG,MAAM,GAAG,IAAI,CAQnE;AAED;;GAEG;AACH,wBAAgB,YAAY,CAC1B,KAAK,EAAE,QAAQ,GAAG,MAAM,EACxB,OAAO,GAAE,aAAa,GAAG;IAAE,MAAM,CAAC,EAAE,MAAM,CAAA;CAAO,GAChD,aAAa,CAmBf"}
|
package/dist/readers/index.js
CHANGED
|
@@ -1,21 +1,50 @@
|
|
|
1
1
|
import { InvalidArgumentError } from "../core/errors.js";
|
|
2
|
+
import { stripCompressionExtension } from "../utils/compression.js";
|
|
3
|
+
import { FileSystemReader } from "../files/reader.js";
|
|
2
4
|
import { CSVReader } from "./csv.js";
|
|
3
5
|
import { JSONReader } from "./json.js";
|
|
4
6
|
import { JSONLReader } from "./jsonl.js";
|
|
7
|
+
import { ParquetReader } from "./parquet.js";
|
|
5
8
|
import { XLSXReader } from "./xlsx.js";
|
|
9
|
+
export * from "../files/reader.js";
|
|
6
10
|
export * from "./csv.js";
|
|
7
11
|
export * from "./json.js";
|
|
8
12
|
export * from "./jsonl.js";
|
|
13
|
+
export * from "./parquet.js";
|
|
9
14
|
export * from "./xlsx.js";
|
|
10
15
|
const adapters = {
|
|
11
16
|
csv: {
|
|
12
17
|
name: "CSV",
|
|
13
|
-
extensions: [".csv", ".
|
|
18
|
+
extensions: [".csv", ".txt"],
|
|
14
19
|
createReader: (input, options) => new CSVReader(input, options),
|
|
15
20
|
createWriter: () => {
|
|
16
21
|
throw new Error("Writer adapter called from reader");
|
|
17
22
|
},
|
|
18
23
|
},
|
|
24
|
+
tsv: {
|
|
25
|
+
name: "TSV",
|
|
26
|
+
extensions: [".tsv", ".tab"],
|
|
27
|
+
createReader: (input, options) => new CSVReader(input, { ...options, delimiter: options?.delimiter ?? "\t" }),
|
|
28
|
+
createWriter: () => {
|
|
29
|
+
throw new Error("Writer adapter called from reader");
|
|
30
|
+
},
|
|
31
|
+
},
|
|
32
|
+
psv: {
|
|
33
|
+
name: "PSV",
|
|
34
|
+
extensions: [".psv"],
|
|
35
|
+
createReader: (input, options) => new CSVReader(input, { ...options, delimiter: options?.delimiter ?? "|" }),
|
|
36
|
+
createWriter: () => {
|
|
37
|
+
throw new Error("Writer adapter called from reader");
|
|
38
|
+
},
|
|
39
|
+
},
|
|
40
|
+
parquet: {
|
|
41
|
+
name: "Parquet",
|
|
42
|
+
extensions: [".parquet", ".pq"],
|
|
43
|
+
createReader: (input, options) => new ParquetReader(input, options),
|
|
44
|
+
createWriter: () => {
|
|
45
|
+
throw new Error("Writer adapter called from reader");
|
|
46
|
+
},
|
|
47
|
+
},
|
|
19
48
|
json: {
|
|
20
49
|
name: "JSON",
|
|
21
50
|
extensions: [".json"],
|
|
@@ -40,12 +69,38 @@ const adapters = {
|
|
|
40
69
|
throw new Error("Writer adapter called from reader");
|
|
41
70
|
},
|
|
42
71
|
},
|
|
72
|
+
files: {
|
|
73
|
+
name: "Filesystem",
|
|
74
|
+
extensions: [],
|
|
75
|
+
createReader: (input, options) => {
|
|
76
|
+
if (typeof input !== "string") {
|
|
77
|
+
throw new InvalidArgumentError("Filesystem reader requires a string path");
|
|
78
|
+
}
|
|
79
|
+
return new FileSystemReader({ root: input, ...options });
|
|
80
|
+
},
|
|
81
|
+
createWriter: () => {
|
|
82
|
+
throw new Error("Filesystem is a read-only source adapter");
|
|
83
|
+
},
|
|
84
|
+
},
|
|
85
|
+
filesystem: {
|
|
86
|
+
name: "Filesystem",
|
|
87
|
+
extensions: [],
|
|
88
|
+
createReader: (input, options) => {
|
|
89
|
+
if (typeof input !== "string") {
|
|
90
|
+
throw new InvalidArgumentError("Filesystem reader requires a string path");
|
|
91
|
+
}
|
|
92
|
+
return new FileSystemReader({ root: input, ...options });
|
|
93
|
+
},
|
|
94
|
+
createWriter: () => {
|
|
95
|
+
throw new Error("Filesystem is a read-only source adapter");
|
|
96
|
+
},
|
|
97
|
+
},
|
|
43
98
|
};
|
|
44
99
|
/**
|
|
45
100
|
* Infers format name from file path or extension.
|
|
46
101
|
*/
|
|
47
102
|
export function inferFormatFromPath(filePath) {
|
|
48
|
-
const cleanPath = filePath.toLowerCase()
|
|
103
|
+
const cleanPath = stripCompressionExtension(filePath.toLowerCase());
|
|
49
104
|
for (const [format, adapter] of Object.entries(adapters)) {
|
|
50
105
|
if (adapter.extensions.some((ext) => cleanPath.endsWith(ext))) {
|
|
51
106
|
return format;
|
|
@@ -62,7 +117,6 @@ export function createReader(input, options = {}) {
|
|
|
62
117
|
format = inferFormatFromPath(input) ?? undefined;
|
|
63
118
|
}
|
|
64
119
|
if (!format) {
|
|
65
|
-
// Default to CSV if cannot infer
|
|
66
120
|
format = "csv";
|
|
67
121
|
}
|
|
68
122
|
const adapter = adapters[format];
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"index.js","sourceRoot":"","sources":["../../src/readers/index.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,oBAAoB,EAAE,MAAM,mBAAmB,CAAC;AAEzD,OAAO,EAAE,SAAS,EAAE,MAAM,UAAU,CAAC;AACrC,OAAO,EAAE,UAAU,EAAE,MAAM,WAAW,CAAC;AACvC,OAAO,EAAE,WAAW,EAAE,MAAM,YAAY,CAAC;AACzC,OAAO,EAAE,UAAU,EAAE,MAAM,WAAW,CAAC;AAEvC,cAAc,UAAU,CAAC;AACzB,cAAc,WAAW,CAAC;AAC1B,cAAc,YAAY,CAAC;AAC3B,cAAc,WAAW,CAAC;AAE1B,MAAM,QAAQ,GAAkC;IAC9C,GAAG,EAAE;QACH,IAAI,EAAE,KAAK;QACX,UAAU,EAAE,CAAC,MAAM,EAAE,MAAM,EAAE,MAAM,CAAC;
|
|
1
|
+
{"version":3,"file":"index.js","sourceRoot":"","sources":["../../src/readers/index.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,oBAAoB,EAAE,MAAM,mBAAmB,CAAC;AAEzD,OAAO,EAAE,yBAAyB,EAAE,MAAM,yBAAyB,CAAC;AACpE,OAAO,EAAE,gBAAgB,EAAE,MAAM,oBAAoB,CAAC;AACtD,OAAO,EAAE,SAAS,EAAE,MAAM,UAAU,CAAC;AACrC,OAAO,EAAE,UAAU,EAAE,MAAM,WAAW,CAAC;AACvC,OAAO,EAAE,WAAW,EAAE,MAAM,YAAY,CAAC;AACzC,OAAO,EAAE,aAAa,EAAE,MAAM,cAAc,CAAC;AAC7C,OAAO,EAAE,UAAU,EAAE,MAAM,WAAW,CAAC;AAEvC,cAAc,oBAAoB,CAAC;AACnC,cAAc,UAAU,CAAC;AACzB,cAAc,WAAW,CAAC;AAC1B,cAAc,YAAY,CAAC;AAC3B,cAAc,cAAc,CAAC;AAC7B,cAAc,WAAW,CAAC;AAE1B,MAAM,QAAQ,GAAkC;IAC9C,GAAG,EAAE;QACH,IAAI,EAAE,KAAK;QACX,UAAU,EAAE,CAAC,MAAM,EAAE,MAAM,CAAC;QAC5B,YAAY,EAAE,CAAC,KAAK,EAAE,OAAO,EAAE,EAAE,CAAC,IAAI,SAAS,CAAC,KAAK,EAAE,OAAO,CAAC;QAC/D,YAAY,EAAE,GAAG,EAAE;YACjB,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;QACvD,CAAC;KACF;IACD,GAAG,EAAE;QACH,IAAI,EAAE,KAAK;QACX,UAAU,EAAE,CAAC,MAAM,EAAE,MAAM,CAAC;QAC5B,YAAY,EAAE,CAAC,KAAK,EAAE,OAAO,EAAE,EAAE,CAC/B,IAAI,SAAS,CAAC,KAAK,EAAE,EAAE,GAAG,OAAO,EAAE,SAAS,EAAE,OAAO,EAAE,SAAS,IAAI,IAAI,EAAE,CAAC;QAC7E,YAAY,EAAE,GAAG,EAAE;YACjB,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;QACvD,CAAC;KACF;IACD,GAAG,EAAE;QACH,IAAI,EAAE,KAAK;QACX,UAAU,EAAE,CAAC,MAAM,CAAC;QACpB,YAAY,EAAE,CAAC,KAAK,EAAE,OAAO,EAAE,EAAE,CAC/B,IAAI,SAAS,CAAC,KAAK,EAAE,EAAE,GAAG,OAAO,EAAE,SAAS,EAAE,OAAO,EAAE,SAAS,IAAI,GAAG,EAAE,CAAC;QAC5E,YAAY,EAAE,GAAG,EAAE;YACjB,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;QACvD,CAAC;KACF;IACD,OAAO,EAAE;QACP,IAAI,EAAE,SAAS;QACf,UAAU,EAAE,CAAC,UAAU,EAAE,KAAK,CAAC;QAC/B,YAAY,EAAE,CAAC,KAAK,EAAE,OAAO,EAAE,EAAE,CAAC,IAAI,aAAa,CAAC,KAAK,EAAE,OAAO,CAAC;QACnE,YAAY,EAAE,GAAG,EAAE;YACjB,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;QACvD,CAAC;KACF;IACD,IAAI,EAAE;QACJ,IAAI,EAAE,MAAM;QACZ,UAAU,EAAE,CAAC,OAAO,CAAC;QACrB,YAAY,EAAE,CAAC,KAAK,EAAE,OAAO,EAAE,EAAE,CAAC,IAAI,UAAU,CAAC,KAAK,EAAE,OAAO,CAAC;QAChE,YAAY,EAAE,GAAG,EAAE;YACjB,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;QACvD,CAAC;KACF;IACD,KAAK,EAAE;QACL,IAAI,EAAE,OAAO;QACb,UAAU,EAAE,CAAC,QAAQ,EAAE,SAAS,EAAE,SAAS,CAAC;QAC5C,YAAY,EAAE,CAAC,KAAK,EAAE,OAAO,EAAE,EAAE,CAAC,IAAI,WAAW,CAAC,KAAK,EAAE,OAAO,CAAC;QACjE,YAAY,EAAE,GAAG,EAAE;YACjB,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;QACvD,CAAC;KACF;IACD,IAAI,EAAE;QACJ,IAAI,EAAE,MAAM;QACZ,UAAU,EAAE,CAAC,OAAO,EAAE,OAAO,CAAC;QAC9B,YAAY,EAAE,CAAC,KAAK,EAAE,OAAO,EAAE,EAAE,CAAC,IAAI,UAAU,CAAC,KAAK,EAAE,OAAO,CAAC;QAChE,YAAY,EAAE,GAAG,EAAE;YACjB,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;QACvD,CAAC;KACF;IACD,KAAK,EAAE;QACL,IAAI,EAAE,YAAY;QAClB,UAAU,EAAE,EAAE;QACd,YAAY,EAAE,CAAC,KAAK,EAAE,OAAO,EAAE,EAAE;YAC/B,IAAI,OAAO,KAAK,KAAK,QAAQ,EAAE,CAAC;gBAC9B,MAAM,IAAI,oBAAoB,CAAC,0CAA0C,CAAC,CAAC;YAC7E,CAAC;YACD,OAAO,IAAI,gBAAgB,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,GAAG,OAAO,EAAE,CAAC,CAAC;QAC3D,CAAC;QACD,YAAY,EAAE,GAAG,EAAE;YACjB,MAAM,IAAI,KAAK,CAAC,0CAA0C,CAAC,CAAC;QAC9D,CAAC;KACF;IACD,UAAU,EAAE;QACV,IAAI,EAAE,YAAY;QAClB,UAAU,EAAE,EAAE;QACd,YAAY,EAAE,CAAC,KAAK,EAAE,OAAO,EAAE,EAAE;YAC/B,IAAI,OAAO,KAAK,KAAK,QAAQ,EAAE,CAAC;gBAC9B,MAAM,IAAI,oBAAoB,CAAC,0CAA0C,CAAC,CAAC;YAC7E,CAAC;YACD,OAAO,IAAI,gBAAgB,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,GAAG,OAAO,EAAE,CAAC,CAAC;QAC3D,CAAC;QACD,YAAY,EAAE,GAAG,EAAE;YACjB,MAAM,IAAI,KAAK,CAAC,0CAA0C,CAAC,CAAC;QAC9D,CAAC;KACF;CACF,CAAC;AAEF;;GAEG;AACH,MAAM,UAAU,mBAAmB,CAAC,QAAgB;IAClD,MAAM,SAAS,GAAG,yBAAyB,CAAC,QAAQ,CAAC,WAAW,EAAE,CAAC,CAAC;IACpE,KAAK,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,QAAQ,CAAC,EAAE,CAAC;QACzD,IAAI,OAAO,CAAC,UAAU,CAAC,IAAI,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,SAAS,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAC,EAAE,CAAC;YAC9D,OAAO,MAAM,CAAC;QAChB,CAAC;IACH,CAAC;IACD,OAAO,IAAI,CAAC;AACd,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,YAAY,CAC1B,KAAwB,EACxB,UAA+C,EAAE;IAEjD,IAAI,MAAM,GAAG,OAAO,CAAC,MAAM,EAAE,WAAW,EAAE,CAAC;IAE3C,IAAI,CAAC,MAAM,IAAI,OAAO,KAAK,KAAK,QAAQ,IAAI,KAAK,KAAK,GAAG,EAAE,CAAC;QAC1D,MAAM,GAAG,mBAAmB,CAAC,KAAK,CAAC,IAAI,SAAS,CAAC;IACnD,CAAC;IAED,IAAI,CAAC,MAAM,EAAE,CAAC;QACZ,MAAM,GAAG,KAAK,CAAC;IACjB,CAAC;IAED,MAAM,OAAO,GAAG,QAAQ,CAAC,MAAM,CAAC,CAAC;IACjC,IAAI,CAAC,OAAO,EAAE,CAAC;QACb,MAAM,IAAI,oBAAoB,CAC5B,8BAA8B,MAAM,6BAA6B,MAAM,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CACpG,CAAC;IACJ,CAAC;IAED,OAAO,OAAO,CAAC,YAAY,CAAC,KAAK,EAAE,OAAO,CAAC,CAAC;AAC9C,CAAC"}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"json.d.ts","sourceRoot":"","sources":["../../src/readers/json.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,aAAa,CAAC;AAE5C,OAAO,KAAK,EAEV,UAAU,EACV,kBAAkB,EAClB,aAAa,EAEb,aAAa,EACd,MAAM,kBAAkB,CAAC;AAE1B,MAAM,WAAW,iBAAkB,SAAQ,aAAa;IACtD,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,QAAQ,CAAC,EAAE,MAAM,CAAC;CACnB;AAED;;;;GAIG;AACH,qBAAa,UAAW,YAAW,aAAa;IAC9C,OAAO,CAAC,KAAK,CAAoB;IACjC,OAAO,CAAC,OAAO,CAAoB;gBAEvB,KAAK,EAAE,QAAQ,GAAG,MAAM,EAAE,OAAO,GAAE,iBAAsB;IAQrE,OAAO,CAAC,cAAc;
|
|
1
|
+
{"version":3,"file":"json.d.ts","sourceRoot":"","sources":["../../src/readers/json.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,aAAa,CAAC;AAE5C,OAAO,KAAK,EAEV,UAAU,EACV,kBAAkB,EAClB,aAAa,EAEb,aAAa,EACd,MAAM,kBAAkB,CAAC;AAE1B,MAAM,WAAW,iBAAkB,SAAQ,aAAa;IACtD,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,QAAQ,CAAC,EAAE,MAAM,CAAC;CACnB;AAED;;;;GAIG;AACH,qBAAa,UAAW,YAAW,aAAa;IAC9C,OAAO,CAAC,KAAK,CAAoB;IACjC,OAAO,CAAC,OAAO,CAAoB;gBAEvB,KAAK,EAAE,QAAQ,GAAG,MAAM,EAAE,OAAO,GAAE,iBAAsB;IAQrE,OAAO,CAAC,cAAc;IAKf,IAAI,CAAC,OAAO,CAAC,EAAE,aAAa,GAAG,UAAU;IAgM1C,OAAO,CAAC,OAAO,CAAC,EAAE,aAAa,GAAG,OAAO,CAAC,kBAAkB,CAAC;CAsBpE"}
|
package/dist/readers/json.js
CHANGED
|
@@ -1,4 +1,4 @@
|
|
|
1
|
-
import {
|
|
1
|
+
import { openDecompressedReadStream } from "../utils/compression.js";
|
|
2
2
|
import { ParseError } from "../core/errors.js";
|
|
3
3
|
/**
|
|
4
4
|
* Incremental streaming JSON parser for top-level arrays [ {...}, ... ]
|
|
@@ -15,11 +15,9 @@ export class JSONReader {
|
|
|
15
15
|
this.options.filePath = input;
|
|
16
16
|
}
|
|
17
17
|
}
|
|
18
|
-
getInputStream() {
|
|
19
|
-
|
|
20
|
-
|
|
21
|
-
}
|
|
22
|
-
return this.input;
|
|
18
|
+
getInputStream(opts) {
|
|
19
|
+
const gzip = opts?.gzip ?? this.options.gzip;
|
|
20
|
+
return openDecompressedReadStream(this.input, { gzip });
|
|
23
21
|
}
|
|
24
22
|
async *read(options) {
|
|
25
23
|
const mergedOptions = {
|
|
@@ -27,8 +25,8 @@ export class JSONReader {
|
|
|
27
25
|
...options,
|
|
28
26
|
};
|
|
29
27
|
const batchSize = Math.max(1, mergedOptions.batchSize ?? 1000);
|
|
30
|
-
const targetPath = mergedOptions.path
|
|
31
|
-
const stream = this.getInputStream();
|
|
28
|
+
const targetPath = mergedOptions.path ?? "";
|
|
29
|
+
const stream = this.getInputStream(mergedOptions);
|
|
32
30
|
let buffer = "";
|
|
33
31
|
let cursor = 0;
|
|
34
32
|
let byteOffset = 0;
|
package/dist/readers/json.js.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"json.js","sourceRoot":"","sources":["../../src/readers/json.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,
|
|
1
|
+
{"version":3,"file":"json.js","sourceRoot":"","sources":["../../src/readers/json.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,0BAA0B,EAAE,MAAM,yBAAyB,CAAC;AAErE,OAAO,EAAE,UAAU,EAAE,MAAM,mBAAmB,CAAC;AAe/C;;;;GAIG;AACH,MAAM,OAAO,UAAU;IACb,KAAK,CAAoB;IACzB,OAAO,CAAoB;IAEnC,YAAY,KAAwB,EAAE,UAA6B,EAAE;QACnE,IAAI,CAAC,KAAK,GAAG,KAAK,CAAC;QACnB,IAAI,CAAC,OAAO,GAAG,EAAE,GAAG,OAAO,EAAE,CAAC;QAC9B,IAAI,OAAO,KAAK,KAAK,QAAQ,EAAE,CAAC;YAC9B,IAAI,CAAC,OAAO,CAAC,QAAQ,GAAG,KAAK,CAAC;QAChC,CAAC;IACH,CAAC;IAEO,cAAc,CAAC,IAAoB;QACzC,MAAM,IAAI,GAAG,IAAI,EAAE,IAAI,IAAI,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC;QAC7C,OAAO,0BAA0B,CAAC,IAAI,CAAC,KAAK,EAAE,EAAE,IAAI,EAAE,CAAC,CAAC;IAC1D,CAAC;IAED,KAAK,CAAC,CAAC,IAAI,CAAC,OAAuB;QACjC,MAAM,aAAa,GAAsB;YACvC,GAAG,IAAI,CAAC,OAAO;YACf,GAAG,OAAO;SACX,CAAC;QACF,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,aAAa,CAAC,SAAS,IAAI,IAAI,CAAC,CAAC;QAC/D,MAAM,UAAU,GAAG,aAAa,CAAC,IAAI,IAAI,EAAE,CAAC;QAC5C,MAAM,MAAM,GAAG,IAAI,CAAC,cAAc,CAAC,aAAa,CAAC,CAAC;QAElD,IAAI,MAAM,GAAG,EAAE,CAAC;QAChB,IAAI,MAAM,GAAG,CAAC,CAAC;QACf,IAAI,UAAU,GAAG,CAAC,CAAC;QACnB,IAAI,gBAAgB,GAAG,CAAC,UAAU,CAAC,CAAC,oCAAoC;QACxE,IAAI,WAAW,GAAa,EAAE,CAAC;QAE/B,wCAAwC;QACxC,IAAI,OAAO,GAAG,KAAK,CAAC;QACpB,IAAI,QAAQ,GAAG,KAAK,CAAC;QACrB,IAAI,SAAS,GAAG,KAAK,CAAC;QACtB,IAAI,WAAW,GAAG,CAAC,CAAC;QACpB,IAAI,gBAAgB,GAAG,CAAC,CAAC,CAAC;QAE1B,IAAI,kBAAkB,GAAU,EAAE,CAAC;QACnC,IAAI,YAAY,GAAG,CAAC,CAAC;QACrB,IAAI,QAAQ,GAAG,CAAC,CAAC;QAEjB,kDAAkD;QAClD,MAAM,eAAe,GAAG,UAAU,CAAC,CAAC,CAAC,UAAU,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;QAEhE,IAAI,KAAK,EAAE,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;YACjC,MAAM,QAAQ,GACZ,OAAO,KAAK,KAAK,QAAQ,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAE,KAAgB,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC;YACzE,MAAM,IAAI,QAAQ,CAAC;YAEnB,OAAO,MAAM,GAAG,MAAM,CAAC,MAAM,EAAE,CAAC;gBAC9B,MAAM,IAAI,GAAG,MAAM,CAAC,MAAM,CAAC,CAAC;gBAE5B,IAAI,SAAS,EAAE,CAAC;oBACd,SAAS,GAAG,KAAK,CAAC;oBAClB,MAAM,EAAE,CAAC;oBACT,UAAU,EAAE,CAAC;oBACb,SAAS;gBACX,CAAC;gBAED,IAAI,IAAI,KAAK,IAAI,EAAE,CAAC;oBAClB,SAAS,GAAG,IAAI,CAAC;oBACjB,MAAM,EAAE,CAAC;oBACT,UAAU,EAAE,CAAC;oBACb,SAAS;gBACX,CAAC;gBAED,IAAI,IAAI,KAAK,GAAG,EAAE,CAAC;oBACjB,QAAQ,GAAG,CAAC,QAAQ,CAAC;oBACrB,MAAM,EAAE,CAAC;oBACT,UAAU,EAAE,CAAC;oBACb,SAAS;gBACX,CAAC;gBAED,IAAI,QAAQ,EAAE,CAAC;oBACb,MAAM,EAAE,CAAC;oBACT,UAAU,EAAE,CAAC;oBACb,SAAS;gBACX,CAAC;gBAED,uCAAuC;gBACvC,IAAI,CAAC,gBAAgB,EAAE,CAAC;oBACtB,qDAAqD;oBACrD,qCAAqC;oBACrC,IAAI,IAAI,KAAK,GAAG,EAAE,CAAC;wBACjB,MAAM,EAAE,CAAC;wBACT,UAAU,EAAE,CAAC;wBACb,SAAS;oBACX,CAAC;yBAAM,IAAI,IAAI,KAAK,GAAG,EAAE,CAAC;wBACxB,2CAA2C;wBAC3C,0CAA0C;wBAC1C,gBAAgB,GAAG,IAAI,CAAC;wBACxB,OAAO,GAAG,IAAI,CAAC;wBACf,MAAM,EAAE,CAAC;wBACT,UAAU,EAAE,CAAC;wBACb,SAAS;oBACX,CAAC;yBAAM,IAAI,IAAI,KAAK,GAAG,EAAE,CAAC;wBACxB,sDAAsD;wBACtD,IAAI,MAAM,GAAG,MAAM,GAAG,CAAC,CAAC;wBACxB,OAAO,MAAM,IAAI,CAAC,IAAI,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,MAAM,CAAE,CAAC;4BAAE,MAAM,EAAE,CAAC;wBAC3D,IAAI,MAAM,IAAI,CAAC,IAAI,MAAM,CAAC,MAAM,CAAC,KAAK,GAAG,EAAE,CAAC;4BAC1C,IAAI,QAAQ,GAAG,MAAM,GAAG,CAAC,CAAC;4BAC1B,OAAO,QAAQ,IAAI,CAAC,IAAI,MAAM,CAAC,QAAQ,CAAC,KAAK,GAAG;gCAAE,QAAQ,EAAE,CAAC;4BAC7D,IAAI,QAAQ,IAAI,CAAC,EAAE,CAAC;gCAClB,MAAM,OAAO,GAAG,MAAM,CAAC,KAAK,CAAC,QAAQ,GAAG,CAAC,EAAE,MAAM,CAAC,CAAC;gCACnD,WAAW,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;gCAC1B,IACE,WAAW,CAAC,IAAI,CAAC,GAAG,CAAC,KAAK,UAAU;oCACpC,OAAO,KAAK,eAAe,CAAC,eAAe,CAAC,MAAM,GAAG,CAAC,CAAC,EACvD,CAAC;oCACD,oCAAoC;oCACpC,8BAA8B;gCAChC,CAAC;4BACH,CAAC;wBACH,CAAC;wBACD,MAAM,EAAE,CAAC;wBACT,UAAU,EAAE,CAAC;wBACb,SAAS;oBACX,CAAC;oBACD,MAAM,EAAE,CAAC;oBACT,UAAU,EAAE,CAAC;oBACb,SAAS;gBACX,CAAC;gBAED,+CAA+C;gBAC/C,IAAI,CAAC,OAAO,EAAE,CAAC;oBACb,IAAI,IAAI,KAAK,GAAG,EAAE,CAAC;wBACjB,OAAO,GAAG,IAAI,CAAC;wBACf,MAAM,EAAE,CAAC;wBACT,UAAU,EAAE,CAAC;wBACb,SAAS;oBACX,CAAC;gBACH,CAAC;qBAAM,CAAC;oBACN,8CAA8C;oBAC9C,IAAI,IAAI,KAAK,GAAG,EAAE,CAAC;wBACjB,IAAI,WAAW,KAAK,CAAC,EAAE,CAAC;4BACtB,gBAAgB,GAAG,MAAM,CAAC;wBAC5B,CAAC;wBACD,WAAW,EAAE,CAAC;oBAChB,CAAC;yBAAM,IAAI,IAAI,KAAK,GAAG,EAAE,CAAC;wBACxB,WAAW,EAAE,CAAC;wBACd,IAAI,WAAW,KAAK,CAAC,IAAI,gBAAgB,KAAK,CAAC,CAAC,EAAE,CAAC;4BACjD,2CAA2C;4BAC3C,MAAM,SAAS,GAAG,MAAM,CAAC,KAAK,CAAC,gBAAgB,EAAE,MAAM,GAAG,CAAC,CAAC,CAAC;4BAC7D,IAAI,SAAkB,CAAC;4BACvB,IAAI,CAAC;gCACH,SAAS,GAAG,IAAI,CAAC,KAAK,CAAC,SAAS,CAAC,CAAC;4BACpC,CAAC;4BAAC,OAAO,GAAG,EAAE,CAAC;gCACb,MAAM,IAAI,UAAU,CAClB,mCAAoC,GAAa,CAAC,OAAO,EAAE,EAC3D;oCACE,IAAI,EAAE,aAAa,CAAC,QAAQ;oCAC5B,GAAG,EAAE,QAAQ,GAAG,CAAC;oCACjB,UAAU;iCACX,CACF,CAAC;4BACJ,CAAC;4BAED,IACE,OAAO,SAAS,KAAK,QAAQ;gCAC7B,SAAS,KAAK,IAAI;gCAClB,CAAC,KAAK,CAAC,OAAO,CAAC,SAAS,CAAC,EACzB,CAAC;gCACD,kBAAkB,CAAC,IAAI,CAAC,SAAgB,CAAC,CAAC;gCAC1C,QAAQ,EAAE,CAAC;gCAEX,IAAI,kBAAkB,CAAC,MAAM,IAAI,SAAS,EAAE,CAAC;oCAC3C,MAAM;wCACJ,IAAI,EAAE,kBAAkB;wCACxB,MAAM,EAAE,YAAY;qCACrB,CAAC;oCACF,YAAY,IAAI,kBAAkB,CAAC,MAAM,CAAC;oCAC1C,kBAAkB,GAAG,EAAE,CAAC;gCAC1B,CAAC;4BACH,CAAC;4BAED,gBAAgB,GAAG,CAAC,CAAC,CAAC;wBACxB,CAAC;oBACH,CAAC;yBAAM,IAAI,IAAI,KAAK,GAAG,IAAI,WAAW,KAAK,CAAC,EAAE,CAAC;wBAC7C,uBAAuB;wBACvB,OAAO,GAAG,KAAK,CAAC;wBAChB,MAAM;oBACR,CAAC;oBAED,MAAM,EAAE,CAAC;oBACT,UAAU,EAAE,CAAC;gBACf,CAAC;YACH,CAAC;YAED,4EAA4E;YAC5E,IAAI,gBAAgB,KAAK,CAAC,CAAC,EAAE,CAAC;gBAC5B,MAAM,GAAG,MAAM,CAAC,KAAK,CAAC,gBAAgB,CAAC,CAAC;gBACxC,MAAM,GAAG,MAAM,GAAG,gBAAgB,CAAC;gBACnC,gBAAgB,GAAG,CAAC,CAAC;YACvB,CAAC;iBAAM,CAAC;gBACN,MAAM,GAAG,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC;gBAC9B,MAAM,GAAG,CAAC,CAAC;YACb,CAAC;QACH,CAAC;QAED,IAAI,kBAAkB,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YAClC,MAAM;gBACJ,IAAI,EAAE,kBAAkB;gBACxB,MAAM,EAAE,YAAY;aACrB,CAAC;QACJ,CAAC;IACH,CAAC;IAED,KAAK,CAAC,OAAO,CAAC,OAAuB;QACnC,IAAI,QAAQ,GAAG,CAAC,CAAC;QACjB,IAAI,WAAW,GAAa,EAAE,CAAC;QAE/B,IAAI,KAAK,EAAE,MAAM,KAAK,IAAI,IAAI,CAAC,IAAI,CAAC,EAAE,GAAG,OAAO,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,EAAE,CAAC;YACrE,IAAI,WAAW,CAAC,MAAM,KAAK,CAAC,IAAI,KAAK,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,IAAI,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC;gBACvE,WAAW,GAAG,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;YAC3C,CAAC;YACD,QAAQ,IAAI,KAAK,CAAC,IAAI,CAAC,MAAM,CAAC;QAChC,CAAC;QAED,OAAO;YACL,MAAM,EAAE,MAAM;YACd,QAAQ;YACR,WAAW,EAAE,WAAW,CAAC,MAAM;YAC/B,OAAO,EAAE,WAAW,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC;gBAClC,IAAI;gBACJ,IAAI,EAAE,QAAQ;gBACd,cAAc,EAAE,CAAC;aAClB,CAAC,CAAC;SACJ,CAAC;IACJ,CAAC;CACF"}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"jsonl.d.ts","sourceRoot":"","sources":["../../src/readers/jsonl.ts"],"names":[],"mappings":"AAEA,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,aAAa,CAAC;AAE5C,OAAO,KAAK,EAEV,UAAU,EACV,kBAAkB,EAClB,aAAa,EAEb,aAAa,EACd,MAAM,kBAAkB,CAAC;AAE1B,MAAM,WAAW,kBAAmB,SAAQ,aAAa;IACvD,QAAQ,CAAC,EAAE,MAAM,CAAC;CACnB;AAED,qBAAa,WAAY,YAAW,aAAa;IAC/C,OAAO,CAAC,KAAK,CAAoB;IACjC,OAAO,CAAC,OAAO,CAAqB;gBAExB,KAAK,EAAE,QAAQ,GAAG,MAAM,EAAE,OAAO,GAAE,kBAAuB;IAQtE,OAAO,CAAC,cAAc;
|
|
1
|
+
{"version":3,"file":"jsonl.d.ts","sourceRoot":"","sources":["../../src/readers/jsonl.ts"],"names":[],"mappings":"AAEA,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,aAAa,CAAC;AAE5C,OAAO,KAAK,EAEV,UAAU,EACV,kBAAkB,EAClB,aAAa,EAEb,aAAa,EACd,MAAM,kBAAkB,CAAC;AAE1B,MAAM,WAAW,kBAAmB,SAAQ,aAAa;IACvD,QAAQ,CAAC,EAAE,MAAM,CAAC;CACnB;AAED,qBAAa,WAAY,YAAW,aAAa;IAC/C,OAAO,CAAC,KAAK,CAAoB;IACjC,OAAO,CAAC,OAAO,CAAqB;gBAExB,KAAK,EAAE,QAAQ,GAAG,MAAM,EAAE,OAAO,GAAE,kBAAuB;IAQtE,OAAO,CAAC,cAAc;IAKf,IAAI,CAAC,OAAO,CAAC,EAAE,aAAa,GAAG,UAAU;IAmE1C,OAAO,CAAC,OAAO,CAAC,EAAE,aAAa,GAAG,OAAO,CAAC,kBAAkB,CAAC;CAsBpE"}
|
package/dist/readers/jsonl.js
CHANGED
|
@@ -1,4 +1,4 @@
|
|
|
1
|
-
import {
|
|
1
|
+
import { openDecompressedReadStream } from "../utils/compression.js";
|
|
2
2
|
import { createInterface } from "node:readline";
|
|
3
3
|
import { ParseError } from "../core/errors.js";
|
|
4
4
|
export class JSONLReader {
|
|
@@ -11,11 +11,9 @@ export class JSONLReader {
|
|
|
11
11
|
this.options.filePath = input;
|
|
12
12
|
}
|
|
13
13
|
}
|
|
14
|
-
getInputStream() {
|
|
15
|
-
|
|
16
|
-
|
|
17
|
-
}
|
|
18
|
-
return this.input;
|
|
14
|
+
getInputStream(opts) {
|
|
15
|
+
const gzip = opts?.gzip ?? this.options.gzip;
|
|
16
|
+
return openDecompressedReadStream(this.input, { gzip });
|
|
19
17
|
}
|
|
20
18
|
async *read(options) {
|
|
21
19
|
const mergedOptions = {
|
|
@@ -23,7 +21,7 @@ export class JSONLReader {
|
|
|
23
21
|
...options,
|
|
24
22
|
};
|
|
25
23
|
const batchSize = Math.max(1, mergedOptions.batchSize ?? 1000);
|
|
26
|
-
const stream = this.getInputStream();
|
|
24
|
+
const stream = this.getInputStream(mergedOptions);
|
|
27
25
|
const rl = createInterface({
|
|
28
26
|
input: stream,
|
|
29
27
|
crlfDelay: Infinity,
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"jsonl.js","sourceRoot":"","sources":["../../src/readers/jsonl.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,
|
|
1
|
+
{"version":3,"file":"jsonl.js","sourceRoot":"","sources":["../../src/readers/jsonl.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,0BAA0B,EAAE,MAAM,yBAAyB,CAAC;AACrE,OAAO,EAAE,eAAe,EAAE,MAAM,eAAe,CAAC;AAEhD,OAAO,EAAE,UAAU,EAAE,MAAM,mBAAmB,CAAC;AAc/C,MAAM,OAAO,WAAW;IACd,KAAK,CAAoB;IACzB,OAAO,CAAqB;IAEpC,YAAY,KAAwB,EAAE,UAA8B,EAAE;QACpE,IAAI,CAAC,KAAK,GAAG,KAAK,CAAC;QACnB,IAAI,CAAC,OAAO,GAAG,EAAE,GAAG,OAAO,EAAE,CAAC;QAC9B,IAAI,OAAO,KAAK,KAAK,QAAQ,EAAE,CAAC;YAC9B,IAAI,CAAC,OAAO,CAAC,QAAQ,GAAG,KAAK,CAAC;QAChC,CAAC;IACH,CAAC;IAEO,cAAc,CAAC,IAAoB;QACzC,MAAM,IAAI,GAAG,IAAI,EAAE,IAAI,IAAI,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC;QAC7C,OAAO,0BAA0B,CAAC,IAAI,CAAC,KAAK,EAAE,EAAE,IAAI,EAAE,CAAC,CAAC;IAC1D,CAAC;IAED,KAAK,CAAC,CAAC,IAAI,CAAC,OAAuB;QACjC,MAAM,aAAa,GAAuB;YACxC,GAAG,IAAI,CAAC,OAAO;YACf,GAAG,OAAO;SACX,CAAC;QACF,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,aAAa,CAAC,SAAS,IAAI,IAAI,CAAC,CAAC;QAC/D,MAAM,MAAM,GAAG,IAAI,CAAC,cAAc,CAAC,aAAa,CAAC,CAAC;QAElD,MAAM,EAAE,GAAG,eAAe,CAAC;YACzB,KAAK,EAAE,MAAM;YACb,SAAS,EAAE,QAAQ;SACpB,CAAC,CAAC;QAEH,IAAI,UAAU,GAAG,CAAC,CAAC;QACnB,IAAI,kBAAkB,GAAU,EAAE,CAAC;QACnC,IAAI,YAAY,GAAG,CAAC,CAAC;QAErB,IAAI,KAAK,EAAE,MAAM,IAAI,IAAI,EAAE,EAAE,CAAC;YAC5B,UAAU,EAAE,CAAC;YACb,MAAM,OAAO,GAAG,IAAI,CAAC,IAAI,EAAE,CAAC;YAC5B,IAAI,CAAC,OAAO;gBAAE,SAAS;YAEvB,IAAI,MAAe,CAAC;YACpB,IAAI,CAAC;gBACH,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC;YAC/B,CAAC;YAAC,OAAO,GAAG,EAAE,CAAC;gBACb,MAAM,IAAI,UAAU,CAClB,wBAAwB,UAAU,KAAM,GAAa,CAAC,OAAO,EAAE,EAC/D;oBACE,IAAI,EAAE,aAAa,CAAC,QAAQ;oBAC5B,IAAI,EAAE,UAAU;oBAChB,GAAG,EAAE,UAAU;iBAChB,CACF,CAAC;YACJ,CAAC;YAED,IAAI,OAAO,MAAM,KAAK,QAAQ,IAAI,MAAM,KAAK,IAAI,IAAI,KAAK,CAAC,OAAO,CAAC,MAAM,CAAC,EAAE,CAAC;gBAC3E,MAAM,IAAI,UAAU,CAClB,gCAAgC,UAAU,SAAS,KAAK,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,OAAO,MAAM,EAAE,EACpG;oBACE,IAAI,EAAE,aAAa,CAAC,QAAQ;oBAC5B,IAAI,EAAE,UAAU;oBAChB,GAAG,EAAE,UAAU;iBAChB,CACF,CAAC;YACJ,CAAC;YAED,kBAAkB,CAAC,IAAI,CAAC,MAAa,CAAC,CAAC;YAEvC,IAAI,kBAAkB,CAAC,MAAM,IAAI,SAAS,EAAE,CAAC;gBAC3C,MAAM;oBACJ,IAAI,EAAE,kBAAkB;oBACxB,MAAM,EAAE,YAAY;iBACrB,CAAC;gBACF,YAAY,IAAI,kBAAkB,CAAC,MAAM,CAAC;gBAC1C,kBAAkB,GAAG,EAAE,CAAC;YAC1B,CAAC;QACH,CAAC;QAED,IAAI,kBAAkB,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YAClC,MAAM;gBACJ,IAAI,EAAE,kBAAkB;gBACxB,MAAM,EAAE,YAAY;aACrB,CAAC;QACJ,CAAC;IACH,CAAC;IAED,KAAK,CAAC,OAAO,CAAC,OAAuB;QACnC,IAAI,QAAQ,GAAG,CAAC,CAAC;QACjB,IAAI,WAAW,GAAa,EAAE,CAAC;QAE/B,IAAI,KAAK,EAAE,MAAM,KAAK,IAAI,IAAI,CAAC,IAAI,CAAC,EAAE,GAAG,OAAO,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,EAAE,CAAC;YACrE,IAAI,WAAW,CAAC,MAAM,KAAK,CAAC,IAAI,KAAK,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,IAAI,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC;gBACvE,WAAW,GAAG,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;YAC3C,CAAC;YACD,QAAQ,IAAI,KAAK,CAAC,IAAI,CAAC,MAAM,CAAC;QAChC,CAAC;QAED,OAAO;YACL,MAAM,EAAE,OAAO;YACf,QAAQ;YACR,WAAW,EAAE,WAAW,CAAC,MAAM;YAC/B,OAAO,EAAE,WAAW,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC;gBAClC,IAAI;gBACJ,IAAI,EAAE,QAAQ;gBACd,cAAc,EAAE,CAAC;aAClB,CAAC,CAAC;SACJ,CAAC;IACJ,CAAC;CACF"}
|
|
@@ -0,0 +1,17 @@
|
|
|
1
|
+
import type { Readable } from "node:stream";
|
|
2
|
+
import type { DataStream, InspectionMetadata, ReaderOptions, TabularReader } from "../core/types.js";
|
|
3
|
+
export interface ParquetReaderOptions extends ReaderOptions {
|
|
4
|
+
filePath?: string;
|
|
5
|
+
}
|
|
6
|
+
/**
|
|
7
|
+
* High-performance pure-JS Parquet Reader supporting Snappy, Zstd, Gzip, and Brotli decompression.
|
|
8
|
+
*/
|
|
9
|
+
export declare class ParquetReader implements TabularReader {
|
|
10
|
+
private input;
|
|
11
|
+
private options;
|
|
12
|
+
constructor(input: Readable | string, options?: ParquetReaderOptions);
|
|
13
|
+
private getArrayBuffer;
|
|
14
|
+
read(options?: ReaderOptions): DataStream;
|
|
15
|
+
inspect(options?: ReaderOptions): Promise<InspectionMetadata>;
|
|
16
|
+
}
|
|
17
|
+
//# sourceMappingURL=parquet.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"parquet.d.ts","sourceRoot":"","sources":["../../src/readers/parquet.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,aAAa,CAAC;AAI5C,OAAO,KAAK,EAGV,UAAU,EACV,kBAAkB,EAClB,aAAa,EAEb,aAAa,EACd,MAAM,kBAAkB,CAAC;AAE1B,MAAM,WAAW,oBAAqB,SAAQ,aAAa;IACzD,QAAQ,CAAC,EAAE,MAAM,CAAC;CACnB;AAgDD;;GAEG;AACH,qBAAa,aAAc,YAAW,aAAa;IACjD,OAAO,CAAC,KAAK,CAAoB;IACjC,OAAO,CAAC,OAAO,CAAuB;gBAE1B,KAAK,EAAE,QAAQ,GAAG,MAAM,EAAE,OAAO,GAAE,oBAAyB;YAQ1D,cAAc;IAerB,IAAI,CAAC,OAAO,CAAC,EAAE,aAAa,GAAG,UAAU;IAyD1C,OAAO,CAAC,OAAO,CAAC,EAAE,aAAa,GAAG,OAAO,CAAC,kBAAkB,CAAC;CAoBpE"}
|
|
@@ -0,0 +1,146 @@
|
|
|
1
|
+
import { readFile } from "node:fs/promises";
|
|
2
|
+
import { parquetMetadataAsync, parquetReadObjects } from "hyparquet";
|
|
3
|
+
import { compressors } from "hyparquet-compressors";
|
|
4
|
+
import { ParseError } from "../core/errors.js";
|
|
5
|
+
/**
|
|
6
|
+
* Maps Parquet primitive and logical types to Rowpipe ColumnType.
|
|
7
|
+
*/
|
|
8
|
+
function mapParquetType(parquetType) {
|
|
9
|
+
if (!parquetType)
|
|
10
|
+
return "mixed";
|
|
11
|
+
switch (parquetType) {
|
|
12
|
+
case "BOOLEAN":
|
|
13
|
+
return "boolean";
|
|
14
|
+
case "INT32":
|
|
15
|
+
case "INT64":
|
|
16
|
+
return "integer";
|
|
17
|
+
case "FLOAT":
|
|
18
|
+
case "DOUBLE":
|
|
19
|
+
return "number";
|
|
20
|
+
case "BYTE_ARRAY":
|
|
21
|
+
case "FIXED_LEN_BYTE_ARRAY":
|
|
22
|
+
case "UTF8":
|
|
23
|
+
return "string";
|
|
24
|
+
case "INT96":
|
|
25
|
+
case "DATE":
|
|
26
|
+
case "TIME_MILLIS":
|
|
27
|
+
case "TIME_MICROS":
|
|
28
|
+
case "TIMESTAMP_MILLIS":
|
|
29
|
+
case "TIMESTAMP_MICROS":
|
|
30
|
+
return "datetime";
|
|
31
|
+
default:
|
|
32
|
+
return "mixed";
|
|
33
|
+
}
|
|
34
|
+
}
|
|
35
|
+
/**
|
|
36
|
+
* Normalizes values returned by Parquet reader (e.g. converting BigInt to Number).
|
|
37
|
+
*/
|
|
38
|
+
function normalizeParquetValue(val) {
|
|
39
|
+
if (typeof val === "bigint") {
|
|
40
|
+
if (val <= BigInt(Number.MAX_SAFE_INTEGER) && val >= BigInt(Number.MIN_SAFE_INTEGER)) {
|
|
41
|
+
return Number(val);
|
|
42
|
+
}
|
|
43
|
+
return Number(val);
|
|
44
|
+
}
|
|
45
|
+
if (val instanceof Uint8Array) {
|
|
46
|
+
return new TextDecoder().decode(val);
|
|
47
|
+
}
|
|
48
|
+
return val;
|
|
49
|
+
}
|
|
50
|
+
/**
|
|
51
|
+
* High-performance pure-JS Parquet Reader supporting Snappy, Zstd, Gzip, and Brotli decompression.
|
|
52
|
+
*/
|
|
53
|
+
export class ParquetReader {
|
|
54
|
+
input;
|
|
55
|
+
options;
|
|
56
|
+
constructor(input, options = {}) {
|
|
57
|
+
this.input = input;
|
|
58
|
+
this.options = { ...options };
|
|
59
|
+
if (typeof input === "string") {
|
|
60
|
+
this.options.filePath = input;
|
|
61
|
+
}
|
|
62
|
+
}
|
|
63
|
+
async getArrayBuffer() {
|
|
64
|
+
if (typeof this.input === "string") {
|
|
65
|
+
const buf = await readFile(this.input);
|
|
66
|
+
return buf.buffer.slice(buf.byteOffset, buf.byteOffset + buf.byteLength);
|
|
67
|
+
}
|
|
68
|
+
// Readable stream: collect chunks into single Buffer
|
|
69
|
+
const chunks = [];
|
|
70
|
+
for await (const chunk of this.input) {
|
|
71
|
+
chunks.push(typeof chunk === "string" ? Buffer.from(chunk) : chunk);
|
|
72
|
+
}
|
|
73
|
+
const combined = Buffer.concat(chunks);
|
|
74
|
+
return combined.buffer.slice(combined.byteOffset, combined.byteOffset + combined.byteLength);
|
|
75
|
+
}
|
|
76
|
+
async *read(options) {
|
|
77
|
+
const mergedOptions = {
|
|
78
|
+
...this.options,
|
|
79
|
+
...options,
|
|
80
|
+
};
|
|
81
|
+
const batchSize = Math.max(1, mergedOptions.batchSize ?? 1000);
|
|
82
|
+
let arrayBuffer;
|
|
83
|
+
try {
|
|
84
|
+
arrayBuffer = await this.getArrayBuffer();
|
|
85
|
+
}
|
|
86
|
+
catch (err) {
|
|
87
|
+
throw new ParseError(`Failed to load Parquet data: ${err.message}`, {
|
|
88
|
+
file: mergedOptions.filePath,
|
|
89
|
+
});
|
|
90
|
+
}
|
|
91
|
+
let rows;
|
|
92
|
+
try {
|
|
93
|
+
rows = (await parquetReadObjects({
|
|
94
|
+
file: arrayBuffer,
|
|
95
|
+
compressors,
|
|
96
|
+
}));
|
|
97
|
+
}
|
|
98
|
+
catch (err) {
|
|
99
|
+
throw new ParseError(`Failed to parse Parquet file: ${err.message}`, {
|
|
100
|
+
file: mergedOptions.filePath,
|
|
101
|
+
});
|
|
102
|
+
}
|
|
103
|
+
let currentBatch = [];
|
|
104
|
+
let globalOffset = 0;
|
|
105
|
+
for (let i = 0; i < rows.length; i++) {
|
|
106
|
+
const rawRow = rows[i];
|
|
107
|
+
const normalizedRow = {};
|
|
108
|
+
for (const [key, val] of Object.entries(rawRow)) {
|
|
109
|
+
normalizedRow[key] = normalizeParquetValue(val);
|
|
110
|
+
}
|
|
111
|
+
currentBatch.push(normalizedRow);
|
|
112
|
+
if (currentBatch.length >= batchSize) {
|
|
113
|
+
yield {
|
|
114
|
+
rows: currentBatch,
|
|
115
|
+
offset: globalOffset,
|
|
116
|
+
};
|
|
117
|
+
globalOffset += currentBatch.length;
|
|
118
|
+
currentBatch = [];
|
|
119
|
+
}
|
|
120
|
+
}
|
|
121
|
+
if (currentBatch.length > 0) {
|
|
122
|
+
yield {
|
|
123
|
+
rows: currentBatch,
|
|
124
|
+
offset: globalOffset,
|
|
125
|
+
};
|
|
126
|
+
}
|
|
127
|
+
}
|
|
128
|
+
async inspect(options) {
|
|
129
|
+
const arrayBuffer = await this.getArrayBuffer();
|
|
130
|
+
const meta = await parquetMetadataAsync(arrayBuffer);
|
|
131
|
+
const rowCount = typeof meta.num_rows === "bigint" ? Number(meta.num_rows) : Number(meta.num_rows || 0);
|
|
132
|
+
const schemaColumns = (meta.schema || []).filter((s) => s.name && s.name !== "root");
|
|
133
|
+
const columns = schemaColumns.map((col) => ({
|
|
134
|
+
name: col.name,
|
|
135
|
+
type: mapParquetType(col.type),
|
|
136
|
+
nullPercentage: 0,
|
|
137
|
+
}));
|
|
138
|
+
return {
|
|
139
|
+
format: "Parquet",
|
|
140
|
+
rowCount,
|
|
141
|
+
columnCount: columns.length,
|
|
142
|
+
columns,
|
|
143
|
+
};
|
|
144
|
+
}
|
|
145
|
+
}
|
|
146
|
+
//# sourceMappingURL=parquet.js.map
|