webforai 3.0.0 → 4.1.0
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 +47 -5
- package/dist/bin.js +3109 -490
- package/dist/cjs/detect-client-shell.js +78 -0
- package/dist/cjs/extract-mdast.js +58 -0
- package/dist/cjs/extract-meta-refresh.js +71 -0
- package/dist/cjs/extractors/index.js +5 -0
- package/dist/cjs/extractors/lib/block-features.js +465 -0
- package/dist/cjs/extractors/lib/block-model.generated.js +45 -0
- package/dist/cjs/extractors/lib/block-model.js +72 -0
- package/dist/cjs/extractors/lib/blocks.js +212 -0
- package/dist/cjs/extractors/lib/code-tabs.js +191 -0
- package/dist/cjs/extractors/lib/constants.js +1 -1
- package/dist/cjs/extractors/lib/layout-tables.js +123 -0
- package/dist/cjs/extractors/lib/sanitize.js +146 -26
- package/dist/cjs/extractors/lib/takumi-signal.js +76 -0
- package/dist/cjs/extractors/lib/ui-chrome.js +7 -0
- package/dist/cjs/extractors/presets/auto.js +3 -3
- package/dist/cjs/extractors/presets/kiwame.js +231 -0
- package/dist/cjs/extractors/presets/takumi.js +34 -3
- package/dist/cjs/html-to-markdown.js +58 -4
- package/dist/cjs/html-to-mdast.js +4 -3
- package/dist/cjs/index.js +16 -0
- package/dist/cjs/link-replacer.js +19 -13
- package/dist/cjs/loaders/cf-puppeteer.js +14 -7
- package/dist/cjs/loaders/fetch.js +14 -2
- package/dist/cjs/loaders/playwright.js +32 -30
- package/dist/cjs/loaders/puppeteer.js +14 -7
- package/dist/cjs/md-splitter.js +24 -11
- package/dist/cjs/mdast-handlers/custom-code-handler.js +32 -9
- package/dist/cjs/mdast-handlers/custom-div-handler.js +14 -35
- package/dist/cjs/mdast-handlers/custom-img-handler.js +10 -1
- package/dist/cjs/mdast-handlers/custom-table-handler.js +18 -1
- package/dist/cjs/mdast-handlers/definition-list-handler.js +17 -6
- package/dist/cjs/mdast-handlers/table-cell-flatten.js +195 -0
- package/dist/cjs/mdast-handlers/twoslash.js +71 -0
- package/dist/cjs/mdast-to-markdown.js +2 -5
- package/dist/cjs/metadata/index.js +4 -3
- package/dist/cjs/normalize/index.js +11 -2
- package/dist/cjs/platform/client.js +73 -16
- package/dist/cjs/platform/index.js +4 -0
- package/dist/cjs/platform/types.js +7 -1
- package/dist/cjs/utils/hast-fast.js +1 -1
- package/dist/cjs/utils/parse-html.js +113 -0
- package/dist/cjs/utils/pre-text.js +104 -0
- package/dist/detect-client-shell.js +53 -0
- package/dist/extract-mdast.js +58 -0
- package/dist/extract-meta-refresh.js +47 -0
- package/dist/extractors/index.js +3 -0
- package/dist/extractors/lib/block-features.js +437 -0
- package/dist/extractors/lib/block-model.generated.js +21 -0
- package/dist/extractors/lib/block-model.js +48 -0
- package/dist/extractors/lib/blocks.js +185 -0
- package/dist/extractors/lib/code-tabs.js +168 -0
- package/dist/extractors/lib/constants.js +1 -1
- package/dist/extractors/lib/layout-tables.js +99 -0
- package/dist/extractors/lib/sanitize.js +144 -26
- package/dist/extractors/lib/takumi-signal.js +53 -0
- package/dist/extractors/lib/ui-chrome.js +5 -0
- package/dist/extractors/presets/auto.js +3 -3
- package/dist/extractors/presets/kiwame.js +206 -0
- package/dist/extractors/presets/takumi.js +34 -3
- package/dist/html-to-markdown.js +57 -4
- package/dist/html-to-mdast.js +4 -3
- package/dist/index.js +18 -0
- package/dist/link-replacer.js +19 -13
- package/dist/loaders/cf-puppeteer.js +15 -8
- package/dist/loaders/fetch.js +14 -2
- package/dist/loaders/playwright.js +32 -30
- package/dist/loaders/puppeteer.js +15 -8
- package/dist/md-splitter.js +24 -11
- package/dist/mdast-handlers/custom-code-handler.js +30 -9
- package/dist/mdast-handlers/custom-div-handler.js +14 -35
- package/dist/mdast-handlers/custom-img-handler.js +10 -1
- package/dist/mdast-handlers/custom-table-handler.js +19 -2
- package/dist/mdast-handlers/definition-list-handler.js +17 -6
- package/dist/mdast-handlers/table-cell-flatten.js +170 -0
- package/dist/mdast-handlers/twoslash.js +48 -0
- package/dist/mdast-to-markdown.js +2 -5
- package/dist/metadata/index.js +4 -3
- package/dist/normalize/index.js +12 -2
- package/dist/platform/client.js +73 -16
- package/dist/platform/index.js +4 -0
- package/dist/platform/types.js +5 -1
- package/dist/types/cli/bin.js.map +1 -1
- package/dist/types/cli/commands/jobs/files.d.ts +28 -0
- package/dist/types/cli/commands/jobs/files.js.map +1 -0
- package/dist/types/cli/commands/jobs/llms-txt.d.ts +33 -0
- package/dist/types/cli/commands/jobs/llms-txt.js.map +1 -0
- package/dist/types/cli/commands/jobs/options.d.ts +58 -0
- package/dist/types/cli/commands/jobs/options.js.map +1 -0
- package/dist/types/cli/commands/jobs/run.d.ts +44 -0
- package/dist/types/cli/commands/jobs/run.js.map +1 -0
- package/dist/types/cli/commands/skill/content.js.map +1 -1
- package/dist/types/cli/commands/webforai/interactive.d.ts +1 -1
- package/dist/types/cli/commands/webforai/interactive.js.map +1 -1
- package/dist/types/cli/commands/webforai/options.d.ts +9 -2
- package/dist/types/cli/commands/webforai/options.js.map +1 -1
- package/dist/types/cli/commands/webforai/run.js.map +1 -1
- package/dist/types/cli/constants.d.ts +3 -1
- package/dist/types/cli/constants.js.map +1 -1
- package/dist/types/cli/utils.d.ts +8 -1
- package/dist/types/cli/utils.js.map +1 -1
- package/dist/types/detect-client-shell.d.ts +24 -0
- package/dist/types/detect-client-shell.js.map +1 -0
- package/dist/types/extract-mdast.js.map +1 -1
- package/dist/types/extract-meta-refresh.d.ts +29 -0
- package/dist/types/extract-meta-refresh.js.map +1 -0
- package/dist/types/extractors/index.d.ts +1 -0
- package/dist/types/extractors/index.js.map +1 -1
- package/dist/types/extractors/lib/block-features.d.ts +43 -0
- package/dist/types/extractors/lib/block-features.js.map +1 -0
- package/dist/types/extractors/lib/block-model.d.ts +28 -0
- package/dist/types/extractors/lib/block-model.generated.d.ts +5 -0
- package/dist/types/extractors/lib/block-model.generated.js.map +1 -0
- package/dist/types/extractors/lib/block-model.js.map +1 -0
- package/dist/types/extractors/lib/blocks.d.ts +53 -0
- package/dist/types/extractors/lib/blocks.js.map +1 -0
- package/dist/types/extractors/lib/code-tabs.d.ts +30 -0
- package/dist/types/extractors/lib/code-tabs.js.map +1 -0
- package/dist/types/extractors/lib/constants.js.map +1 -1
- package/dist/types/extractors/lib/layout-tables.d.ts +17 -0
- package/dist/types/extractors/lib/layout-tables.js.map +1 -0
- package/dist/types/extractors/lib/sanitize.d.ts +33 -3
- package/dist/types/extractors/lib/sanitize.js.map +1 -1
- package/dist/types/extractors/lib/takumi-signal.d.ts +12 -0
- package/dist/types/extractors/lib/takumi-signal.js.map +1 -0
- package/dist/types/extractors/lib/ui-chrome.d.ts +8 -0
- package/dist/types/extractors/lib/ui-chrome.js.map +1 -1
- package/dist/types/extractors/presets/auto.d.ts +6 -1
- package/dist/types/extractors/presets/auto.js.map +1 -1
- package/dist/types/extractors/presets/kiwame.d.ts +31 -0
- package/dist/types/extractors/presets/kiwame.js.map +1 -0
- package/dist/types/extractors/presets/takumi.js.map +1 -1
- package/dist/types/html-to-markdown.d.ts +9 -0
- package/dist/types/html-to-markdown.js.map +1 -1
- package/dist/types/html-to-mdast.js.map +1 -1
- package/dist/types/index.d.ts +3 -1
- package/dist/types/index.js.map +1 -1
- package/dist/types/link-replacer.d.ts +3 -1
- package/dist/types/link-replacer.js.map +1 -1
- package/dist/types/loaders/cf-puppeteer.js.map +1 -1
- package/dist/types/loaders/fetch.d.ts +4 -0
- package/dist/types/loaders/fetch.js.map +1 -1
- package/dist/types/loaders/playwright.js.map +1 -1
- package/dist/types/loaders/puppeteer.js.map +1 -1
- package/dist/types/md-splitter.d.ts +1 -1
- package/dist/types/md-splitter.js.map +1 -1
- package/dist/types/mdast-handlers/custom-code-handler.d.ts +9 -0
- package/dist/types/mdast-handlers/custom-code-handler.js.map +1 -1
- package/dist/types/mdast-handlers/custom-div-handler.js.map +1 -1
- package/dist/types/mdast-handlers/custom-img-handler.js.map +1 -1
- package/dist/types/mdast-handlers/custom-table-handler.js.map +1 -1
- package/dist/types/mdast-handlers/definition-list-handler.d.ts +0 -11
- package/dist/types/mdast-handlers/definition-list-handler.js.map +1 -1
- package/dist/types/mdast-handlers/table-cell-flatten.d.ts +18 -0
- package/dist/types/mdast-handlers/table-cell-flatten.js.map +1 -0
- package/dist/types/mdast-handlers/twoslash.d.ts +14 -0
- package/dist/types/mdast-handlers/twoslash.js.map +1 -0
- package/dist/types/mdast-to-markdown.js.map +1 -1
- package/dist/types/metadata/index.js.map +1 -1
- package/dist/types/normalize/index.js.map +1 -1
- package/dist/types/platform/client.d.ts +2 -1
- package/dist/types/platform/client.js.map +1 -1
- package/dist/types/platform/error.d.ts +5 -3
- package/dist/types/platform/error.js.map +1 -1
- package/dist/types/platform/fetch.d.ts +2 -0
- package/dist/types/platform/fetch.js.map +1 -1
- package/dist/types/platform/index.d.ts +1 -1
- package/dist/types/platform/index.js.map +1 -1
- package/dist/types/platform/types.d.ts +35 -6
- package/dist/types/platform/types.js.map +1 -1
- package/dist/types/utils/hast-fast.d.ts +1 -1
- package/dist/types/utils/hast-fast.js.map +1 -1
- package/dist/types/utils/parse-html.d.ts +33 -0
- package/dist/types/utils/parse-html.js.map +1 -0
- package/dist/types/utils/pre-text.d.ts +12 -0
- package/dist/types/utils/pre-text.js.map +1 -0
- package/dist/utils/hast-fast.js +1 -1
- package/dist/utils/parse-html.js +87 -0
- package/dist/utils/pre-text.js +81 -0
- package/package.json +18 -5
|
@@ -0,0 +1,78 @@
|
|
|
1
|
+
"use strict";
|
|
2
|
+
var __defProp = Object.defineProperty;
|
|
3
|
+
var __getOwnPropDesc = Object.getOwnPropertyDescriptor;
|
|
4
|
+
var __getOwnPropNames = Object.getOwnPropertyNames;
|
|
5
|
+
var __hasOwnProp = Object.prototype.hasOwnProperty;
|
|
6
|
+
var __export = (target, all) => {
|
|
7
|
+
for (var name in all)
|
|
8
|
+
__defProp(target, name, { get: all[name], enumerable: true });
|
|
9
|
+
};
|
|
10
|
+
var __copyProps = (to, from, except, desc) => {
|
|
11
|
+
if (from && typeof from === "object" || typeof from === "function") {
|
|
12
|
+
for (let key of __getOwnPropNames(from))
|
|
13
|
+
if (!__hasOwnProp.call(to, key) && key !== except)
|
|
14
|
+
__defProp(to, key, { get: () => from[key], enumerable: !(desc = __getOwnPropDesc(from, key)) || desc.enumerable });
|
|
15
|
+
}
|
|
16
|
+
return to;
|
|
17
|
+
};
|
|
18
|
+
var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
|
|
19
|
+
var detect_client_shell_exports = {};
|
|
20
|
+
__export(detect_client_shell_exports, {
|
|
21
|
+
SHELL_EMPTY_THRESHOLD: () => SHELL_EMPTY_THRESHOLD,
|
|
22
|
+
SHELL_SUSPICION_THRESHOLD: () => SHELL_SUSPICION_THRESHOLD,
|
|
23
|
+
detectClientShell: () => detectClientShell
|
|
24
|
+
});
|
|
25
|
+
module.exports = __toCommonJS(detect_client_shell_exports);
|
|
26
|
+
const SHELL_SUSPICION_THRESHOLD = 300;
|
|
27
|
+
const SHELL_EMPTY_THRESHOLD = 50;
|
|
28
|
+
const BODY_PATTERN = /<body[^>]*>([\s\S]*?)<\/body>/i;
|
|
29
|
+
const HEAD_PATTERN = /<head[^>]*>[\s\S]*?<\/head>/gi;
|
|
30
|
+
const NOSCRIPT_PATTERN = /<noscript[^>]*>([\s\S]*?)<\/noscript>/gi;
|
|
31
|
+
const INVISIBLE_ELEMENT_PATTERN = /<(script|style|template|svg|noscript)[^>]*>[\s\S]*?<\/\1>/gi;
|
|
32
|
+
const COMMENT_PATTERN = /<!--[\s\S]*?-->/g;
|
|
33
|
+
const TAG_PATTERN = /<[^>]+>/g;
|
|
34
|
+
const MOUNT_POINT_PATTERN = /<[a-z][^>]*\sid=["'](?:root|app|__next|___gatsby|__nuxt|q-app|svelte|react-root)["']|<app-root[\s>]|\sdata-reactroot[\s>=]/i;
|
|
35
|
+
const CHALLENGE_PATTERN = /cf-chl|challenge-platform|cf_chl_opt|just a moment|attention required|checking your browser|_incapsula_|px-captcha|datadome/i;
|
|
36
|
+
const ENTITIES = {
|
|
37
|
+
" ": " ",
|
|
38
|
+
"&": "&",
|
|
39
|
+
"<": "<",
|
|
40
|
+
">": ">",
|
|
41
|
+
""": '"',
|
|
42
|
+
"'": "'"
|
|
43
|
+
};
|
|
44
|
+
const visibleTextOf = (markup) => markup.replace(COMMENT_PATTERN, " ").replace(INVISIBLE_ELEMENT_PATTERN, " ").replace(TAG_PATTERN, " ").replace(/&(?:nbsp|amp|lt|gt|quot|#39);/g, (entity) => ENTITIES[entity] ?? " ").replace(/\s+/g, " ").trim();
|
|
45
|
+
const noscriptMentionsJavascript = (html) => {
|
|
46
|
+
for (const match of html.matchAll(NOSCRIPT_PATTERN)) {
|
|
47
|
+
if (/javascript/i.test(match[1] ?? "")) {
|
|
48
|
+
return true;
|
|
49
|
+
}
|
|
50
|
+
}
|
|
51
|
+
return false;
|
|
52
|
+
};
|
|
53
|
+
const detectClientShell = (html) => {
|
|
54
|
+
const body = BODY_PATTERN.exec(html)?.[1] ?? html.replace(HEAD_PATTERN, " ");
|
|
55
|
+
const visibleTextLength = visibleTextOf(body).length;
|
|
56
|
+
if (visibleTextLength >= SHELL_SUSPICION_THRESHOLD) {
|
|
57
|
+
return { isShell: false, visibleTextLength };
|
|
58
|
+
}
|
|
59
|
+
if (CHALLENGE_PATTERN.test(html)) {
|
|
60
|
+
return { isShell: true, reason: "anti-bot-challenge", visibleTextLength };
|
|
61
|
+
}
|
|
62
|
+
if (MOUNT_POINT_PATTERN.test(body)) {
|
|
63
|
+
return { isShell: true, reason: "spa-shell", visibleTextLength };
|
|
64
|
+
}
|
|
65
|
+
if (noscriptMentionsJavascript(html)) {
|
|
66
|
+
return { isShell: true, reason: "noscript-only", visibleTextLength };
|
|
67
|
+
}
|
|
68
|
+
if (visibleTextLength < SHELL_EMPTY_THRESHOLD) {
|
|
69
|
+
return { isShell: true, reason: "empty-body", visibleTextLength };
|
|
70
|
+
}
|
|
71
|
+
return { isShell: false, visibleTextLength };
|
|
72
|
+
};
|
|
73
|
+
// Annotate the CommonJS export names for ESM import in node:
|
|
74
|
+
0 && (module.exports = {
|
|
75
|
+
SHELL_EMPTY_THRESHOLD,
|
|
76
|
+
SHELL_SUSPICION_THRESHOLD,
|
|
77
|
+
detectClientShell
|
|
78
|
+
});
|
|
@@ -32,6 +32,61 @@ const emptyDeclarationFilter = (node) => {
|
|
|
32
32
|
}
|
|
33
33
|
return true;
|
|
34
34
|
};
|
|
35
|
+
const trimEdge = (children, edge) => {
|
|
36
|
+
const index = edge === "start" ? 0 : children.length - 1;
|
|
37
|
+
const node = children[index];
|
|
38
|
+
if (node?.type !== "text") {
|
|
39
|
+
return children;
|
|
40
|
+
}
|
|
41
|
+
const value = edge === "start" ? node.value.trimStart() : node.value.trimEnd();
|
|
42
|
+
const copy = [...children];
|
|
43
|
+
if (value.length === 0) {
|
|
44
|
+
copy.splice(index, 1);
|
|
45
|
+
} else {
|
|
46
|
+
copy[index] = { ...node, value };
|
|
47
|
+
}
|
|
48
|
+
return copy;
|
|
49
|
+
};
|
|
50
|
+
const liftDisplayMath = (paragraph) => {
|
|
51
|
+
if (!paragraph.children.some((child) => child.type === "math")) {
|
|
52
|
+
return void 0;
|
|
53
|
+
}
|
|
54
|
+
const out = [];
|
|
55
|
+
let run = [];
|
|
56
|
+
const flush = () => {
|
|
57
|
+
const trimmed = trimEdge(trimEdge(run, "start"), "end");
|
|
58
|
+
if (trimmed.length > 0) {
|
|
59
|
+
out.push({ ...paragraph, children: trimmed });
|
|
60
|
+
}
|
|
61
|
+
run = [];
|
|
62
|
+
};
|
|
63
|
+
for (const child of paragraph.children) {
|
|
64
|
+
if (child.type === "math") {
|
|
65
|
+
flush();
|
|
66
|
+
out.push(child);
|
|
67
|
+
} else {
|
|
68
|
+
run.push(child);
|
|
69
|
+
}
|
|
70
|
+
}
|
|
71
|
+
flush();
|
|
72
|
+
return out;
|
|
73
|
+
};
|
|
74
|
+
const liftDisplayMathIn = (node) => {
|
|
75
|
+
if (!("children" in node)) {
|
|
76
|
+
return;
|
|
77
|
+
}
|
|
78
|
+
const parent = node;
|
|
79
|
+
for (let index = 0; index < parent.children.length; index++) {
|
|
80
|
+
const child = parent.children[index];
|
|
81
|
+
const lifted = child.type === "paragraph" ? liftDisplayMath(child) : void 0;
|
|
82
|
+
if (lifted) {
|
|
83
|
+
parent.children.splice(index, 1, ...lifted);
|
|
84
|
+
index += lifted.length - 1;
|
|
85
|
+
continue;
|
|
86
|
+
}
|
|
87
|
+
liftDisplayMathIn(child);
|
|
88
|
+
}
|
|
89
|
+
};
|
|
35
90
|
const extractMdast = (node) => {
|
|
36
91
|
const extracted = (0, import_unist_util_filter.filter)(node, (node2) => {
|
|
37
92
|
if (!emptyDeclarationFilter(node2)) {
|
|
@@ -39,6 +94,9 @@ const extractMdast = (node) => {
|
|
|
39
94
|
}
|
|
40
95
|
return true;
|
|
41
96
|
});
|
|
97
|
+
if (extracted) {
|
|
98
|
+
liftDisplayMathIn(extracted);
|
|
99
|
+
}
|
|
42
100
|
return extracted;
|
|
43
101
|
};
|
|
44
102
|
// Annotate the CommonJS export names for ESM import in node:
|
|
@@ -0,0 +1,71 @@
|
|
|
1
|
+
"use strict";
|
|
2
|
+
var __defProp = Object.defineProperty;
|
|
3
|
+
var __getOwnPropDesc = Object.getOwnPropertyDescriptor;
|
|
4
|
+
var __getOwnPropNames = Object.getOwnPropertyNames;
|
|
5
|
+
var __hasOwnProp = Object.prototype.hasOwnProperty;
|
|
6
|
+
var __export = (target, all) => {
|
|
7
|
+
for (var name in all)
|
|
8
|
+
__defProp(target, name, { get: all[name], enumerable: true });
|
|
9
|
+
};
|
|
10
|
+
var __copyProps = (to, from, except, desc) => {
|
|
11
|
+
if (from && typeof from === "object" || typeof from === "function") {
|
|
12
|
+
for (let key of __getOwnPropNames(from))
|
|
13
|
+
if (!__hasOwnProp.call(to, key) && key !== except)
|
|
14
|
+
__defProp(to, key, { get: () => from[key], enumerable: !(desc = __getOwnPropDesc(from, key)) || desc.enumerable });
|
|
15
|
+
}
|
|
16
|
+
return to;
|
|
17
|
+
};
|
|
18
|
+
var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
|
|
19
|
+
var extract_meta_refresh_exports = {};
|
|
20
|
+
__export(extract_meta_refresh_exports, {
|
|
21
|
+
MAX_META_REFRESH_DELAY_SECONDS: () => MAX_META_REFRESH_DELAY_SECONDS,
|
|
22
|
+
extractMetaRefresh: () => extractMetaRefresh
|
|
23
|
+
});
|
|
24
|
+
module.exports = __toCommonJS(extract_meta_refresh_exports);
|
|
25
|
+
const MAX_META_REFRESH_DELAY_SECONDS = 10;
|
|
26
|
+
const META_TAG_PATTERN = /<meta\b[^>]*>/gi;
|
|
27
|
+
const HTTP_EQUIV_REFRESH_PATTERN = /\bhttp-equiv\s*=\s*("\s*refresh\s*"|'\s*refresh\s*'|refresh\b)/i;
|
|
28
|
+
const CONTENT_ATTR_PATTERN = /\bcontent\s*=\s*(?:"([^"]*)"|'([^']*)'|([^\s>]+))/i;
|
|
29
|
+
const CONTENT_VALUE_PATTERN = /^\s*(\d+(?:\.\d+)?)\s*[;,]\s*url\s*=\s*(?:"([^"]*)"|'([^']*)'|(.+?))\s*$/i;
|
|
30
|
+
const resolveTarget = (raw, baseUrl) => {
|
|
31
|
+
try {
|
|
32
|
+
return baseUrl === void 0 ? new URL(raw) : new URL(raw, baseUrl);
|
|
33
|
+
} catch {
|
|
34
|
+
return void 0;
|
|
35
|
+
}
|
|
36
|
+
};
|
|
37
|
+
const extractMetaRefresh = (html, baseUrl) => {
|
|
38
|
+
for (const [tag] of html.matchAll(META_TAG_PATTERN)) {
|
|
39
|
+
if (!HTTP_EQUIV_REFRESH_PATTERN.test(tag)) {
|
|
40
|
+
continue;
|
|
41
|
+
}
|
|
42
|
+
const contentMatch = CONTENT_ATTR_PATTERN.exec(tag);
|
|
43
|
+
const content = contentMatch?.[1] ?? contentMatch?.[2] ?? contentMatch?.[3];
|
|
44
|
+
if (!content) {
|
|
45
|
+
continue;
|
|
46
|
+
}
|
|
47
|
+
const value = CONTENT_VALUE_PATTERN.exec(content);
|
|
48
|
+
if (!value) {
|
|
49
|
+
continue;
|
|
50
|
+
}
|
|
51
|
+
const delaySeconds = Number.parseFloat(value[1] ?? "");
|
|
52
|
+
const raw = (value[2] ?? value[3] ?? value[4] ?? "").trim();
|
|
53
|
+
if (!raw || delaySeconds > MAX_META_REFRESH_DELAY_SECONDS) {
|
|
54
|
+
continue;
|
|
55
|
+
}
|
|
56
|
+
const target = resolveTarget(raw, baseUrl);
|
|
57
|
+
if (!target || target.protocol !== "http:" && target.protocol !== "https:") {
|
|
58
|
+
continue;
|
|
59
|
+
}
|
|
60
|
+
if (baseUrl !== void 0 && resolveTarget(baseUrl, void 0)?.href === target.href) {
|
|
61
|
+
continue;
|
|
62
|
+
}
|
|
63
|
+
return { url: target.href, delaySeconds };
|
|
64
|
+
}
|
|
65
|
+
return void 0;
|
|
66
|
+
};
|
|
67
|
+
// Annotate the CommonJS export names for ESM import in node:
|
|
68
|
+
0 && (module.exports = {
|
|
69
|
+
MAX_META_REFRESH_DELAY_SECONDS,
|
|
70
|
+
extractMetaRefresh
|
|
71
|
+
});
|
|
@@ -20,6 +20,8 @@ var extractors_exports = {};
|
|
|
20
20
|
__export(extractors_exports, {
|
|
21
21
|
autoExtractor: () => import_auto.autoExtractor,
|
|
22
22
|
createAutoExtractor: () => import_auto.createAutoExtractor,
|
|
23
|
+
createKiwameExtractor: () => import_kiwame.createKiwameExtractor,
|
|
24
|
+
kiwameExtractor: () => import_kiwame.kiwameExtractor,
|
|
23
25
|
minimalFilter: () => import_minimal_filter.minimalFilter,
|
|
24
26
|
pipeExtractors: () => import_pipeExtractors.pipeExtractors,
|
|
25
27
|
takumiExtractor: () => import_takumi.takumiExtractor
|
|
@@ -27,12 +29,15 @@ __export(extractors_exports, {
|
|
|
27
29
|
module.exports = __toCommonJS(extractors_exports);
|
|
28
30
|
var import_pipeExtractors = require("./pipeExtractors");
|
|
29
31
|
var import_takumi = require("./presets/takumi");
|
|
32
|
+
var import_kiwame = require("./presets/kiwame");
|
|
30
33
|
var import_minimal_filter = require("./presets/minimal-filter");
|
|
31
34
|
var import_auto = require("./presets/auto");
|
|
32
35
|
// Annotate the CommonJS export names for ESM import in node:
|
|
33
36
|
0 && (module.exports = {
|
|
34
37
|
autoExtractor,
|
|
35
38
|
createAutoExtractor,
|
|
39
|
+
createKiwameExtractor,
|
|
40
|
+
kiwameExtractor,
|
|
36
41
|
minimalFilter,
|
|
37
42
|
pipeExtractors,
|
|
38
43
|
takumiExtractor
|