webforai 1.5.0 → 1.6.0
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/dist/bin.js +972 -213
- package/dist/cjs/cli/bin.js +33 -0
- package/dist/cjs/cli/commands/webforai/index.js +86 -0
- package/dist/cjs/cli/commands/webforai/loadHtml.js +93 -0
- package/dist/cjs/cli/constants.js +34 -0
- package/dist/cjs/cli/helpers/assertContinue.js +34 -0
- package/dist/cjs/cli/helpers/inputOutputPath.js +89 -0
- package/dist/cjs/cli/helpers/inputSourcePath.js +64 -0
- package/dist/cjs/cli/helpers/selectExtractMode.js +39 -0
- package/dist/cjs/cli/helpers/selectLoader.js +44 -0
- package/dist/cjs/cli/utils.js +89 -0
- package/dist/cjs/extract-hast/index.js +4 -3
- package/dist/cjs/extract-hast/readability.js +28 -61
- package/dist/cjs/extract-hast/strongly.js +2 -1
- package/dist/cjs/extract-hast/utils.js +0 -15
- package/dist/cjs/html-to-mdast.js +6 -6
- package/dist/cjs/link-replacer.js +10 -5
- package/dist/cjs/loaders/playwright.js +36 -7
- package/dist/cjs/mdast-handlers/custom-br-handler.js +31 -0
- package/dist/cjs/mdast-to-markdown.js +1 -1
- package/dist/cjs/utils/hast-utils.js +71 -0
- package/dist/cli/bin.js +11 -0
- package/dist/cli/commands/webforai/index.js +53 -0
- package/dist/cli/commands/webforai/loadHtml.js +60 -0
- package/dist/cli/constants.js +9 -0
- package/dist/cli/helpers/assertContinue.js +11 -0
- package/dist/cli/helpers/inputOutputPath.js +56 -0
- package/dist/cli/helpers/inputSourcePath.js +31 -0
- package/dist/cli/helpers/selectExtractMode.js +16 -0
- package/dist/cli/helpers/selectLoader.js +21 -0
- package/dist/cli/utils.js +52 -0
- package/dist/extract-hast/index.js +4 -3
- package/dist/extract-hast/readability.js +29 -62
- package/dist/extract-hast/strongly.js +2 -1
- package/dist/extract-hast/utils.js +0 -14
- package/dist/html-to-mdast.js +6 -6
- package/dist/link-replacer.js +10 -5
- package/dist/loaders/playwright.js +36 -7
- package/dist/mdast-handlers/custom-br-handler.js +8 -0
- package/dist/mdast-to-markdown.js +1 -1
- package/dist/types/cli/bin.js.map +1 -0
- package/dist/types/cli/commands/webforai/index.d.ts +6 -0
- package/dist/types/cli/commands/webforai/index.js.map +1 -0
- package/dist/types/cli/commands/webforai/loadHtml.d.ts +3 -0
- package/dist/types/cli/commands/webforai/loadHtml.js.map +1 -0
- package/dist/types/cli/constants.d.ts +5 -0
- package/dist/types/cli/constants.js.map +1 -0
- package/dist/types/cli/helpers/assertContinue.d.ts +1 -0
- package/dist/types/cli/helpers/assertContinue.js.map +1 -0
- package/dist/types/cli/helpers/inputOutputPath.d.ts +1 -0
- package/dist/types/cli/helpers/inputOutputPath.js.map +1 -0
- package/dist/types/cli/helpers/inputSourcePath.d.ts +1 -0
- package/dist/types/cli/helpers/inputSourcePath.js.map +1 -0
- package/dist/types/cli/helpers/selectExtractMode.d.ts +1 -0
- package/dist/types/cli/helpers/selectExtractMode.js.map +1 -0
- package/dist/types/cli/helpers/selectLoader.d.ts +1 -0
- package/dist/types/cli/helpers/selectLoader.js.map +1 -0
- package/dist/types/cli/utils.d.ts +32 -0
- package/dist/types/cli/utils.js.map +1 -0
- package/dist/types/extract-hast/index.d.ts +9 -4
- package/dist/types/extract-hast/index.js.map +1 -1
- package/dist/types/extract-hast/readability.d.ts +2 -1
- package/dist/types/extract-hast/readability.js.map +1 -1
- package/dist/types/extract-hast/strongly.d.ts +2 -1
- package/dist/types/extract-hast/strongly.js.map +1 -1
- package/dist/types/extract-hast/utils.d.ts +0 -1
- package/dist/types/extract-hast/utils.js.map +1 -1
- package/dist/types/html-to-mdast.d.ts +4 -0
- package/dist/types/html-to-mdast.js.map +1 -1
- package/dist/types/link-replacer.d.ts +1 -1
- package/dist/types/link-replacer.js.map +1 -1
- package/dist/types/loaders/playwright.d.ts +8 -2
- package/dist/types/loaders/playwright.js.map +1 -1
- package/dist/types/mdast-handlers/custom-br-handler.d.ts +2 -0
- package/dist/types/mdast-handlers/custom-br-handler.js.map +1 -0
- package/dist/types/mdast-to-markdown.js.map +1 -1
- package/dist/types/utils/hast-utils.d.ts +4 -0
- package/dist/types/utils/hast-utils.js.map +1 -0
- package/dist/utils/hast-utils.js +46 -0
- package/package.json +18 -17
- package/dist/cjs/bin.js +0 -254
- package/dist/cjs/html-to-markdown.test.js +0 -146
- package/dist/cjs/link-replacer.test.js +0 -47
- package/dist/cjs/loaders/fetch.test.js +0 -9
- package/dist/cjs/loaders/playwright.test.js +0 -17
- package/dist/cjs/loaders/puppeteer.test.js +0 -17
- package/dist/cjs/utils/bin-utils.js +0 -124
- package/dist/html-to-markdown.test.js +0 -124
- package/dist/link-replacer.test.js +0 -47
- package/dist/loaders/fetch.test.js +0 -9
- package/dist/loaders/playwright.test.js +0 -17
- package/dist/loaders/puppeteer.test.js +0 -17
- package/dist/types/bin.js.map +0 -1
- package/dist/types/utils/bin-utils.d.ts +0 -5
- package/dist/types/utils/bin-utils.js.map +0 -1
- package/dist/utils/bin-utils.js +0 -87
- /package/dist/types/{bin.d.ts → cli/bin.d.ts} +0 -0
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"readability.js","sourceRoot":"","sources":["../../../src/extract-hast/readability.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,MAAM,EAAE,SAAS,EAAE,MAAM,kBAAkB,CAAC;AACrD,OAAO,EAAE,QAAQ,IAAI,YAAY,EAAE,MAAM,qBAAqB,CAAC;AAC/D,OAAO,EAAE,MAAM,EAAE,MAAM,mBAAmB,CAAC;
|
|
1
|
+
{"version":3,"file":"readability.js","sourceRoot":"","sources":["../../../src/extract-hast/readability.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,MAAM,EAAE,SAAS,EAAE,MAAM,kBAAkB,CAAC;AACrD,OAAO,EAAE,QAAQ,IAAI,YAAY,EAAE,MAAM,qBAAqB,CAAC;AAC/D,OAAO,EAAE,MAAM,EAAE,MAAM,mBAAmB,CAAC;AAE3C,OAAO,EAAE,UAAU,EAAE,YAAY,EAAE,WAAW,EAAE,MAAM,SAAS,CAAC;AAEhE,MAAM,cAAc,GAAG,CAAC,MAAM,EAAE,SAAS,EAAE,eAAe,EAAE,YAAY,EAAE,OAAO,EAAE,aAAa,EAAE,QAAQ,CAAC,CAAC;AAE5G,MAAM,OAAO,GAAG;IACf,MAAM,EAAE,kCAAkC;IAC1C,MAAM,EAAE,4CAA4C;IACpD,yBAAyB,EAAE,8BAA8B;IACzD,kBAAkB,EACjB,wRAAwR;IACzR,oBAAoB,EAAE,mDAAmD;CACzE,CAAC;AAEF,MAAM,cAAc,GAAG,CAAC,SAAS,EAAE,UAAU,EAAE,eAAe,EAAE,eAAe,EAAE,UAAU,EAAE,QAAQ,CAAC,CAAC;AAEvG,MAAM,cAAc,GAAG,CAAC,GAAG,EAAE,GAAG,EAAE,KAAK,EAAE,SAAS,EAAE,SAAS,EAAE,MAAM,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,CAAC,CAAC;AAEzF,MAAM,mBAAmB,GAAG;IAC3B,EAAE,EAAE,GAAG;IACP,EAAE,EAAE,GAAG;CACP,CAAC;AAEF,yBAAyB;AACzB,UAAU;AACV,WAAW;AACX,KAAK;AAEL,MAAM,cAAc,GAAG,CAAC,IAAU,EAAE,EAAE;IACrC,OAAO,CAAC,CACP,CAAC,SAAS,EAAE,SAAS,CAAC,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC;QAC1C,CAAC,IAAI,CAAC,IAAI,KAAK,SAAS,IAAI,CAAC,QAAQ,EAAE,OAAO,EAAE,MAAM,EAAE,MAAM,EAAE,UAAU,EAAE,KAAK,EAAE,OAAO,CAAC,CAAC,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CACnH,CAAC;AACH,CAAC,CAAC;AAEF,MAAM,sBAAsB,GAAG,CAAC,IAAU,EAAE,EAAE;IAC7C,IAAI,IAAI,CAAC,IAAI,KAAK,SAAS,EAAE,CAAC;QAC7B,OAAO,IAAI,CAAC;IACb,CAAC;IACD,MAAM,OAAO,GAAG,IAAe,CAAC;IAEhC,IAAI,CAAC,OAAO,EAAE,KAAK,CAAC,CAAC,QAAQ,CAAC,OAAO,CAAC,OAAO,CAAC,EAAE,CAAC;QAChD,OAAO,KAAK,CAAC;IACd,CAAC;IAED,yCAAyC;IACzC,IAAI,CAAC,QAAQ,EAAE,aAAa,CAAC,CAAC,IAAI,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,OAAO,CAAC,UAAU,CAAC,GAAG,CAAC,CAAC,EAAE,CAAC;QACtE,OAAO,KAAK,CAAC;IACd,CAAC;IACD,IAAI,UAAU,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,CAAC,SAAS,EAAE,EAAE,CAAC,OAAO,CAAC,MAAM,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC,EAAE,CAAC;QAC7E,OAAO,KAAK,CAAC;IACd,CAAC;IAED,yBAAyB;IACzB,IAAI,OAAO,CAAC,OAAO,KAAK,QAAQ,EAAE,CAAC;QAClC,OAAO,KAAK,CAAC;IACd,CAAC;IACD,IAAI,OAAO,CAAC,UAAU,CAAC,IAAI,KAAK,QAAQ,IAAI,OAAO,CAAC,UAAU,CAAC,YAAY,CAAC,EAAE,CAAC;QAC9E,OAAO,KAAK,CAAC;IACd,CAAC;IAED,yBAAyB;IACzB,IAAI,OAAO,CAAC,UAAU,CAAC,GAAG,KAAK,QAAQ,IAAI,YAAY,CAAC,OAAO,CAAC,UAAU,CAAC,QAAQ,EAAE,QAAQ,CAAC,EAAE,CAAC;QAChG,OAAO,KAAK,CAAC;IACd,CAAC;IACD,IAAI,OAAO,CAAC,MAAM,CAAC,IAAI,CAAC,WAAW,CAAC,OAAO,CAAC,CAAC,EAAE,CAAC;QAC/C,OAAO,KAAK,CAAC;IACd,CAAC;IAED,wBAAwB;IACxB,IAAI,OAAO,CAAC,UAAU,CAAC,IAAI,IAAI,cAAc,CAAC,QAAQ,CAAC,OAAO,CAAC,UAAU,CAAC,IAAc,CAAC,EAAE,CAAC;QAC3F,OAAO,KAAK,CAAC;IACd,CAAC;IAED,OAAO,IAAI,CAAC;AACb,CAAC,CAAC;AAEF,MAAM,qBAAqB,GAAG,CAAC,IAAU,EAAE,EAAE;IAC5C,IAAI,IAAI,CAAC,IAAI,KAAK,SAAS,EAAE,CAAC;QAC7B,OAAO,IAAI,CAAC;IACb,CAAC;IACD,MAAM,OAAO,GAAG,IAAe,CAAC;IAEhC,6BAA6B;IAC7B,IAAI,CAAC,MAAM,EAAE,SAAS,EAAE,MAAM,EAAE,SAAS,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,GAAG,CAAC,CAAC,QAAQ,CAAC,OAAO,CAAC,OAAO,CAAC,EAAE,CAAC;QAC/G,OAAO,IAAI,CAAC;IACb,CAAC;IACD,MAAM,KAAK,GAAG,WAAW,CAAC,OAAO,CAAC,CAAC;IAEnC,IAAI,OAAO,CAAC,yBAAyB,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC;QACnD,OAAO,KAAK,CAAC;IACd,CAAC;IAED,6BAA6B;IAC7B,IAAI,OAAO,CAAC,kBAAkB,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,oBAAoB,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC;QACzF,OAAO,KAAK,CAAC;IACd,CAAC;IAED,OAAO,IAAI,CAAC;AACb,CAAC,CAAC;AAEF,MAAM,iBAAiB,GAAG,CAAC,IAAU,EAAE,KAAa,EAAE,EAAE;IACvD,IAAI,IAAI,CAAC,IAAI,KAAK,SAAS,EAAE,CAAC;QAC7B,OAAO,IAAI,CAAC;IACb,CAAC;IACD,MAAM,OAAO,GAAG,IAAe,CAAC;IAEhC,IAAI,cAAc,CAAC,QAAQ,CAAC,OAAO,CAAC,OAAO,CAAC,EAAE,CAAC;QAC9C,OAAO,IAAI,CAAC;IACb,CAAC;IAED,IAAI,OAAO,CAAC,OAAO,KAAK,KAAK,IAAI,CAAC,OAAO,CAAC,UAAU,CAAC,GAAG,EAAE,CAAC;QAC1D,OAAO,KAAK,CAAC;IACd,CAAC;IAED,OAAO,IAAI,CAAC;AACb,CAAC,CAAC;AAEF,MAAM,CAAC,MAAM,sBAAsB,GAAG,CAAC,MAAqB,EAAQ,EAAE;IACrE,MAAM,EAAE,IAAI,EAAE,IAAI,GAAG,IAAI,EAAE,GAAG,MAAM,CAAC;IACrC,MAAM,IAAI,GAAG,MAAM,CAAC,MAAM,EAAE,IAAI,CAAC,IAAI,IAAI,CAAC;IAE1C,MAAM,oBAAoB,GAAG,MAAM,CAAC,IAAI,EAAE,CAAC,IAAI,EAAE,EAAE,CAAC,cAAc,CAAC,IAAY,CAAC,CAAC,CAAC;IAClF,MAAM,wBAAwB,GAAG,oBAAoB,IAAI,YAAY,CAAC,oBAAoB,CAAC,CAAC;IAC5F,IAAI,CAAC,CAAC,oBAAoB,IAAI,wBAAwB,CAAC,EAAE,CAAC;QACzD,OAAO,IAAI,CAAC;IACb,CAAC;IAED,MAAM,WAAW,GAAG,MAAM,CAAC,oBAAoB,EAAE,CAAC,IAAI,EAAE,EAAE,CAAC,sBAAsB,CAAC,IAAY,CAAC,CAAC,CAAC;IACjG,MAAM,eAAe,GAAG,WAAW,CAAC,CAAC,CAAC,YAAY,CAAC,WAAW,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;IACrE,MAAM,CAAC,QAAQ,EAAE,QAAQ,CAAC,GACzB,eAAe,CAAC,MAAM,GAAG,wBAAwB,CAAC,MAAM,GAAG,CAAC,IAAI,eAAe,CAAC,MAAM,GAAG,IAAI;QAC5F,CAAC,CAAE,CAAC,WAAmB,EAAE,eAAe,CAAW;QACnD,CAAC,CAAE,CAAC,oBAA4B,EAAE,wBAAwB,CAAW,CAAC;IAExE,IAAI,aAAa,GAAG,IAAI,IAAI,mBAAmB,CAAC,CAAC,CAAC,mBAAmB,CAAC,IAAwC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC;IACtH,IAAI,QAAQ,CAAC,MAAM,GAAG,aAAa,EAAE,CAAC;QACrC,aAAa,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,QAAQ,CAAC,MAAM,GAAG,GAAG,CAAC,CAAC;IACpD,CAAC;IAED,IAAI,aAAa,GAAS,QAAQ,CAAC;IACnC,IAAI,aAAa,GAAG,QAAQ,CAAC;IAE7B,KAAK,MAAM,QAAQ,IAAI,cAAc,EAAE,CAAC;QACvC,MAAM,OAAO,GAAG,EAAE,IAAI,EAAE,MAAe,EAAE,QAAQ,EAAE,SAAS,CAAC,QAAQ,EAAE,WAAW,CAAC,EAAE,CAAC;QACtF,MAAM,WAAW,GAAG,YAAY,CAAC,OAAO,CAAC,CAAC;QAE1C,IAAI,WAAW,CAAC,MAAM,GAAG,EAAE,EAAE,CAAC;YAC7B,SAAS;QACV,CAAC;QAED,MAAM,KAAK,GAAG,SAAS,CAAC,GAAG,EAAE,OAAO,CAAC,CAAC;QACtC,MAAM,QAAQ,GAAG,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,YAAY,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;QAElE,MAAM,WAAW,GAAG,QAAQ,CAAC,MAAM,GAAG,WAAW,CAAC,MAAM,CAAC;QACzD,IAAI,WAAW,GAAG,GAAG,EAAE,CAAC;YACvB,SAAS;QACV,CAAC;QAED,IAAI,WAAW,CAAC,MAAM,GAAG,aAAa,EAAE,CAAC;YACxC,aAAa,GAAG,OAAO,CAAC;YACxB,aAAa,GAAG,WAAW,CAAC;YAC5B,MAAM;QACP,CAAC;IACF,CAAC;IAED,MAAM,iBAAiB,GAAG,MAAM,CAAC,aAAa,EAAE,CAAC,IAAI,EAAE,EAAE;QACxD,IAAI,CAAC,iBAAiB,CAAC,IAAY,EAAE,IAAI,CAAC,EAAE,CAAC;YAC5C,OAAO,KAAK,CAAC;QACd,CAAC;QACD,IAAI,CAAC,qBAAqB,CAAC,IAAY,CAAC,EAAE,CAAC;YAC1C,OAAO,KAAK,CAAC;QACd,CAAC;QAED,OAAO,IAAI,CAAC;IACb,CAAC,CAAS,CAAC;IAEX,MAAM,SAAS,GACd,YAAY,CAAC,iBAAiB,CAAC,CAAC,MAAM,GAAG,aAAa,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,iBAAiB,CAAC,CAAC,CAAC,aAAa,CAAC;IACvG,OAAO,SAAS,CAAC;AAClB,CAAC,CAAC"}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"strongly.js","sourceRoot":"","sources":["../../../src/extract-hast/strongly.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,MAAM,EAAE,MAAM,kBAAkB,CAAC;AAC1C,OAAO,EAAE,MAAM,EAAE,MAAM,mBAAmB,CAAC;
|
|
1
|
+
{"version":3,"file":"strongly.js","sourceRoot":"","sources":["../../../src/extract-hast/strongly.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,MAAM,EAAE,MAAM,kBAAkB,CAAC;AAC1C,OAAO,EAAE,MAAM,EAAE,MAAM,mBAAmB,CAAC;AAG3C,MAAM,YAAY,GACjB,iPAAiP,CAAC;AAEnP,MAAM,CAAC,MAAM,mBAAmB,GAAG,CAAC,MAAqB,EAAQ,EAAE;IAClE,MAAM,EAAE,IAAI,EAAE,GAAG,MAAM,CAAC;IACxB,MAAM,SAAS,GAAG,CAAC,SAAS,EAAE,UAAU,EAAE,UAAU,EAAE,MAAM,EAAE,eAAe,CAAC,CAAC;IAC/E,IAAI,eAAe,GAAG,IAAI,CAAC;IAC3B,KAAK,MAAM,QAAQ,IAAI,SAAS,EAAE,CAAC;QAClC,MAAM,OAAO,GAAG,MAAM,CAAC,QAAQ,EAAE,IAAI,CAAC,CAAC;QACvC,IAAI,OAAO,EAAE,CAAC;YACb,eAAe,GAAG,OAAO,CAAC;YAC1B,MAAM;QACP,CAAC;IACF,CAAC;IAED,MAAM,eAAe,GAAG,MAAM,CAAC,eAAe,EAAE,CAAC,IAAI,EAAE,EAAE;QACxD,IAAI,IAAI,CAAC,IAAI,KAAK,SAAS,EAAE,CAAC;YAC7B,OAAO,KAAK,CAAC;QACd,CAAC;QACD,IAAI,IAAI,CAAC,IAAI,KAAK,SAAS,EAAE,CAAC;YAC7B,OAAO,IAAI,CAAC;QACb,CAAC;QACD,MAAM,OAAO,GAAG,IAAe,CAAC;QAChC,MAAM,WAAW,GAAG,GAAG,OAAO,CAAC,OAAO,IAAI,OAAO,CAAC,UAAU,CAAC,EAAE,IAAI,OAAO,CAAC,UAAU,CAAC,SAAS,EAAE,CAAC;QAClG,IAAI,YAAY,CAAC,IAAI,CAAC,WAAW,CAAC,EAAE,CAAC;YACpC,OAAO,KAAK,CAAC;QACd,CAAC;QACD,OAAO,IAAI,CAAC;IACb,CAAC,CAAC,CAAC;IAEH,OAAO,eAAuB,CAAC;AAChC,CAAC,CAAC"}
|
|
@@ -2,4 +2,3 @@ import type { Element } from "hast";
|
|
|
2
2
|
export declare const matchString: (element: Element) => string;
|
|
3
3
|
export declare const classnames: (element: Element) => string[];
|
|
4
4
|
export declare const isStrInclude: (value: unknown, match: string) => boolean;
|
|
5
|
-
export declare const hasAncestors: (element: Element, tagNames: string[], depth?: number) => boolean;
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"utils.js","sourceRoot":"","sources":["../../../src/extract-hast/utils.ts"],"names":[],"mappings":"AAEA,MAAM,CAAC,MAAM,WAAW,GAAG,CAAC,OAAgB,EAAE,EAAE,CAC/C,GAAG,OAAO,CAAC,OAAO,IAAI,OAAO,CAAC,UAAU,CAAC,EAAE,IAAI,UAAU,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC;AAEhF,MAAM,CAAC,MAAM,UAAU,GAAG,CAAC,OAAgB,EAAE,EAAE;IAC9C,IAAI,KAAK,CAAC,OAAO,CAAC,OAAO,CAAC,UAAU,CAAC,SAAS,CAAC,EAAE,CAAC;QACjD,OAAO,OAAO,CAAC,UAAU,CAAC,SAAqB,CAAC;IACjD,CAAC;IACD,OAAO,EAAE,CAAC;AACX,CAAC,CAAC;AAEF,MAAM,CAAC,MAAM,YAAY,GAAG,CAAC,KAAc,EAAE,KAAa,EAAE,EAAE;IAC7D,IAAI,OAAO,KAAK,KAAK,QAAQ,EAAE,CAAC;QAC/B,OAAO,KAAK,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC;IAC9B,CAAC;IACD,OAAO,KAAK,CAAC;AACd,CAAC,CAAC
|
|
1
|
+
{"version":3,"file":"utils.js","sourceRoot":"","sources":["../../../src/extract-hast/utils.ts"],"names":[],"mappings":"AAEA,MAAM,CAAC,MAAM,WAAW,GAAG,CAAC,OAAgB,EAAE,EAAE,CAC/C,GAAG,OAAO,CAAC,OAAO,IAAI,OAAO,CAAC,UAAU,CAAC,EAAE,IAAI,UAAU,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC;AAEhF,MAAM,CAAC,MAAM,UAAU,GAAG,CAAC,OAAgB,EAAE,EAAE;IAC9C,IAAI,KAAK,CAAC,OAAO,CAAC,OAAO,CAAC,UAAU,CAAC,SAAS,CAAC,EAAE,CAAC;QACjD,OAAO,OAAO,CAAC,UAAU,CAAC,SAAqB,CAAC;IACjD,CAAC;IACD,OAAO,EAAE,CAAC;AACX,CAAC,CAAC;AAEF,MAAM,CAAC,MAAM,YAAY,GAAG,CAAC,KAAc,EAAE,KAAa,EAAE,EAAE;IAC7D,IAAI,OAAO,KAAK,KAAK,QAAQ,EAAE,CAAC;QAC/B,OAAO,KAAK,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC;IAC9B,CAAC;IACD,OAAO,KAAK,CAAC;AACd,CAAC,CAAC"}
|
|
@@ -13,6 +13,10 @@ export type HtmlToMdastOptions = {
|
|
|
13
13
|
tableAsText?: boolean;
|
|
14
14
|
/** Whether to hide images. */
|
|
15
15
|
hideImage?: boolean;
|
|
16
|
+
/** The language of the HTML. */
|
|
17
|
+
lang?: string;
|
|
18
|
+
/** The URL of the HTML. */
|
|
19
|
+
url?: string;
|
|
16
20
|
};
|
|
17
21
|
/**
|
|
18
22
|
* Converts an HTML string or HAST tree to an MDAST tree.
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"html-to-mdast.js","sourceRoot":"","sources":["../../src/html-to-mdast.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,QAAQ,EAAE,MAAM,qBAAqB,CAAC;AAC/C,OAAO,EAAE,OAAO,EAAE,MAAM,oBAAoB,CAAC;AAG7C,OAAO,EAAmB,WAAW,EAAE,MAAM,gBAAgB,CAAC;AAC9D,OAAO,EAAE,YAAY,EAAE,MAAM,iBAAiB,CAAC;AAC/C,OAAO,EAAE,cAAc,EAAE,MAAM,mCAAmC,CAAC;AACnE,OAAO,EAAE,iBAAiB,EAAE,MAAM,sCAAsC,CAAC;AACzE,OAAO,EAAE,gBAAgB,EAAE,MAAM,qCAAqC,CAAC;AACvE,OAAO,EAAE,gBAAgB,EAAE,MAAM,qCAAqC,CAAC;AACvE,OAAO,EAAE,kBAAkB,EAAE,MAAM,uCAAuC,CAAC;AAC3E,OAAO,EAAE,
|
|
1
|
+
{"version":3,"file":"html-to-mdast.js","sourceRoot":"","sources":["../../src/html-to-mdast.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,QAAQ,EAAE,MAAM,qBAAqB,CAAC;AAC/C,OAAO,EAAE,OAAO,EAAE,MAAM,oBAAoB,CAAC;AAG7C,OAAO,EAAmB,WAAW,EAAE,MAAM,gBAAgB,CAAC;AAC9D,OAAO,EAAE,YAAY,EAAE,MAAM,iBAAiB,CAAC;AAC/C,OAAO,EAAE,cAAc,EAAE,MAAM,mCAAmC,CAAC;AACnE,OAAO,EAAE,iBAAiB,EAAE,MAAM,sCAAsC,CAAC;AACzE,OAAO,EAAE,gBAAgB,EAAE,MAAM,qCAAqC,CAAC;AACvE,OAAO,EAAE,gBAAgB,EAAE,MAAM,qCAAqC,CAAC;AACvE,OAAO,EAAE,kBAAkB,EAAE,MAAM,uCAAuC,CAAC;AAC3E,OAAO,EAAE,WAAW,EAAE,MAAM,+BAA+B,CAAC;AAC5D,OAAO,EAAE,eAAe,EAAE,cAAc,EAAE,cAAc,EAAE,MAAM,oBAAoB,CAAC;AAoBrF;;;;;;;;;;;;;;;;GAgBG;AACH,MAAM,CAAC,MAAM,WAAW,GAAG,CAAC,UAAyB,EAAE,OAA4B,EAAS,EAAE;IAC7F,MAAM,EAAE,UAAU,EAAE,GAAG,EAAE,UAAU,EAAE,IAAI,EAAE,WAAW,EAAE,GAAG,OAAO,IAAI,EAAE,CAAC;IAEzE,MAAM,CAAC,IAAI,EAAE,IAAI,CAAC,GACjB,OAAO,UAAU,KAAK,QAAQ;QAC7B,CAAC,CAAC,CAAC,WAAW,IAAI,cAAc,CAAC,UAAU,CAAC,EAAE,QAAQ,CAAC,UAAU,EAAE,EAAE,QAAQ,EAAE,IAAI,EAAE,CAAC,CAAC;QACvF,CAAC,CAAC,CAAC,WAAW,IAAI,eAAe,CAAC,UAAU,CAAC,EAAE,UAAU,CAAC,CAAC;IAE7D,MAAM,GAAG,GAAG,UAAU,IAAI,cAAc,CAAC,IAAI,CAAC,CAAC;IAE/C,MAAM,aAAa,GAAG,WAAW,CAAC,EAAE,IAAI,EAAE,IAAI,EAAE,GAAG,EAAE,EAAE,UAAU,CAAC,CAAC;IAEnE,MAAM,KAAK,GAAG,OAAO,CAAC,aAAa,EAAE;QACpC,QAAQ,EAAE;YACT,IAAI,EAAE,WAAW;YACjB,GAAG,EAAE,gBAAgB;YACrB,GAAG,EAAE,iBAAiB;YACtB,CAAC,EAAE,cAAc,CAAC,EAAE,MAAM,EAAE,OAAO,EAAE,UAAU,EAAE,CAAC;YAClD,GAAG,EAAE,gBAAgB,CAAC,EAAE,SAAS,EAAE,OAAO,EAAE,SAAS,EAAE,CAAC;YACxD,KAAK,EAAE,kBAAkB,CAAC,EAAE,MAAM,EAAE,OAAO,EAAE,WAAW,EAAE,CAAC;SAC3D;KACD,CAAC,CAAC;IAEH,MAAM,cAAc,GAAG,YAAY,CAAC,KAAK,CAAC,CAAC;IAE3C,OAAO,cAAc,CAAC;AACvB,CAAC,CAAC"}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
export declare const linkReplacer: (markdown: string,
|
|
1
|
+
export declare const linkReplacer: (markdown: string, base: string) => string;
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"link-replacer.js","sourceRoot":"","sources":["../../src/link-replacer.ts"],"names":[],"mappings":"AAAA,MAAM,CAAC,MAAM,YAAY,GAAG,CAAC,QAAgB,EAAE,IAAY,EAAE,EAAE;IAC9D,MAAM,KAAK,GAAG,
|
|
1
|
+
{"version":3,"file":"link-replacer.js","sourceRoot":"","sources":["../../src/link-replacer.ts"],"names":[],"mappings":"AAAA,MAAM,CAAC,MAAM,YAAY,GAAG,CAAC,QAAgB,EAAE,IAAY,EAAE,EAAE;IAC9D,MAAM,KAAK,GAAG,6BAA6B,CAAC;IAC5C,OAAO,QAAQ,CAAC,OAAO,CAAC,KAAK,EAAE,CAAC,KAAK,EAAE,GAAG,EAAE,GAAG,EAAE,IAAI,EAAE,EAAE;QACxD,IAAI,cAAc,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC;YAC9B,OAAO,KAAK,CAAC;QACd,CAAC;QACD,IAAI,CAAC;YACJ,MAAM,WAAW,GAAG,IAAI,GAAG,CAAC,GAAG,EAAE,IAAI,CAAC,CAAC,IAAI,CAAC;YAC5C,OAAO,GAAG,GAAG,GAAG,WAAW,GAAG,IAAI,EAAE,CAAC;QACtC,CAAC;QAAC,MAAM,CAAC;YACR,OAAO,KAAK,CAAC;QACd,CAAC;IACF,CAAC,CAAC,CAAC;AACJ,CAAC,CAAC"}
|
|
@@ -1,4 +1,10 @@
|
|
|
1
|
-
import { type Browser
|
|
1
|
+
import { type Browser } from "playwright-core";
|
|
2
|
+
export type LoadHtmlOptions = {
|
|
3
|
+
browser?: Browser;
|
|
4
|
+
timeout?: number;
|
|
5
|
+
waitUntil?: "load" | "domcontentloaded" | "networkidle";
|
|
6
|
+
superBypassMode?: boolean;
|
|
7
|
+
};
|
|
2
8
|
/**
|
|
3
9
|
* Useful function for load the HTML of a URL using Playwright.
|
|
4
10
|
* **Not recommended** for use in production environments.
|
|
@@ -13,4 +19,4 @@ import { type Browser, type BrowserContext } from "playwright";
|
|
|
13
19
|
* console.log(html);
|
|
14
20
|
* ```
|
|
15
21
|
*/
|
|
16
|
-
export declare const loadHtml: (url: string,
|
|
22
|
+
export declare const loadHtml: (url: string, options?: LoadHtmlOptions) => Promise<string>;
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"playwright.js","sourceRoot":"","sources":["../../../src/loaders/playwright.ts"],"names":[],"mappings":"AAAA,OAAO,
|
|
1
|
+
{"version":3,"file":"playwright.js","sourceRoot":"","sources":["../../../src/loaders/playwright.ts"],"names":[],"mappings":"AAAA,OAAO,EAAgB,QAAQ,EAAE,OAAO,EAAE,MAAM,iBAAiB,CAAC;AASlE,MAAM,mBAAmB,GAAG;IAC3B,SAAS,EACR,qHAAqH;IACtH,QAAQ,EAAE,EAAE,KAAK,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE;IACvC,iBAAiB,EAAE,CAAC;IACpB,QAAQ,EAAE,KAAK;IACf,QAAQ,EAAE,KAAK;IACf,iBAAiB,EAAE,IAAI;IACvB,MAAM,EAAE,OAAO;IACf,UAAU,EAAE,kBAAkB;CAC9B,CAAC;AAEF;;;;;;;;;;;;;GAaG;AACH,MAAM,CAAC,MAAM,QAAQ,GAAG,KAAK,EAAE,GAAW,EAAE,OAAyB,EAAE,EAAE;IACxE,MAAM,EAAE,OAAO,EAAE,SAAS,EAAE,OAAO,EAAE,eAAe,EAAE,GAAG,OAAO,IAAI,EAAE,CAAC;IACvE,MAAM,QAAQ,GAAG,OAAO,IAAI,CAAC,MAAM,QAAQ,CAAC,MAAM,CAAC,EAAE,QAAQ,EAAE,IAAI,EAAE,CAAC,CAAC,CAAC;IACxE,MAAM,OAAO,GAAG,MAAM,QAAQ,CAAC,UAAU,CAAC,eAAe,CAAC,CAAC,CAAC,mBAAmB,CAAC,CAAC,CAAC,OAAO,CAAC,gBAAgB,CAAC,CAAC,CAAC;IAE7G,IAAI,eAAe,EAAE,CAAC;QACrB,MAAM,OAAO,CAAC,aAAa,CAAC,GAAG,EAAE;YAChC,MAAM,CAAC,cAAc,CAAC,SAAS,EAAE,WAAW,EAAE,EAAE,GAAG,EAAE,GAAG,EAAE,CAAC,SAAS,EAAE,CAAC,CAAC;YACxE,MAAM,CAAC,cAAc,CAAC,SAAS,EAAE,WAAW,EAAE,EAAE,GAAG,EAAE,GAAG,EAAE,CAAC,CAAC,OAAO,EAAE,IAAI,CAAC,EAAE,CAAC,CAAC;YAC9E,MAAM,CAAC,cAAc,CAAC,SAAS,EAAE,SAAS,EAAE,EAAE,GAAG,EAAE,GAAG,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC;QAC7E,CAAC,CAAC,CAAC;IACJ,CAAC;IAED,MAAM,IAAI,GAAG,MAAM,OAAO,CAAC,OAAO,EAAE,CAAC;IACrC,IAAI,eAAe,EAAE,CAAC;QACrB,MAAM,IAAI,CAAC,KAAK,CAAC,SAAS,EAAE,CAAC,KAAK,EAAE,EAAE;YACrC,IAAI,KAAK,CAAC,OAAO,EAAE,CAAC,GAAG,EAAE,CAAC,QAAQ,CAAC,kBAAkB,CAAC,EAAE,CAAC;gBACxD,OAAO,KAAK,CAAC,KAAK,EAAE,CAAC;YACtB,CAAC;YACD,OAAO,KAAK,CAAC,QAAQ,EAAE,CAAC;QACzB,CAAC,CAAC,CAAC;IACJ,CAAC;IAED,MAAM,IAAI,CAAC,IAAI,CAAC,GAAG,EAAE,EAAE,SAAS,EAAE,SAAS,IAAI,MAAM,EAAE,OAAO,EAAE,CAAC,CAAC;IAClE,MAAM,IAAI,GAAG,MAAM,IAAI,CAAC,OAAO,EAAE,CAAC;IAClC,MAAM,IAAI,CAAC,KAAK,EAAE,CAAC;IAEnB,IAAI,OAAO,EAAE,CAAC;QACb,MAAM,OAAO,CAAC,KAAK,EAAE,CAAC;IACvB,CAAC;SAAM,CAAC;QACP,MAAM,QAAQ,CAAC,KAAK,EAAE,CAAC;IACxB,CAAC;IAED,OAAO,IAAI,CAAC;AACb,CAAC,CAAC"}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"custom-br-handler.js","sourceRoot":"","sources":["../../../src/mdast-handlers/custom-br-handler.ts"],"names":[],"mappings":"AAAA,OAAO,EAAe,eAAe,EAAE,MAAM,oBAAoB,CAAC;AAElE,MAAM,CAAC,MAAM,eAAe,GAAW,CAAC,KAAK,EAAE,IAAI,EAAE,EAAE;IACtD,OAAO,eAAe,CAAC,EAAE,CAAC,KAAK,EAAE,IAAI,CAAC,CAAC;AACxC,CAAC,CAAC"}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"mdast-to-markdown.js","sourceRoot":"","sources":["../../src/mdast-to-markdown.ts"],"names":[],"mappings":"AAEA,OAAO,EAAE,aAAa,EAAE,MAAM,gBAAgB,CAAC;AAC/C,OAAO,EAAE,cAAc,EAAE,MAAM,iBAAiB,CAAC;AACjD,OAAO,EAAqC,UAAU,EAAE,MAAM,wBAAwB,CAAC;AAEvF,OAAO,EAAE,YAAY,EAAE,MAAM,iBAAiB,CAAC;AAC/C,OAAO,EAAE,QAAQ,EAAE,MAAM,qBAAqB,CAAC;AAY/C;;GAEG;AACH,MAAM,CAAC,MAAM,iCAAiC,GAA2B;IACxE,UAAU,EAAE,CAAC,aAAa,EAAE,EAAE,cAAc,EAAE,CAAC;IAC/C,MAAM,EAAE,GAAG;CACX,CAAC;AAEF;;;;;;;;;;;;;;;;;;;;;;;;;;GA0BG;AACH,MAAM,CAAC,MAAM,eAAe,GAAG,CAAC,KAA4B,EAAE,OAAgC,EAAU,EAAE;IACzG,MAAM,EAAE,OAAO,EAAE,GAAG,iBAAiB,EAAE,GAAG,EAAE,GAAG,iCAAiC,EAAE,GAAG,OAAO,EAAE,CAAC;IAE/F,IAAI,QAAQ,GAAG,UAAU,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,iBAAiB,CAAC,CAAC;
|
|
1
|
+
{"version":3,"file":"mdast-to-markdown.js","sourceRoot":"","sources":["../../src/mdast-to-markdown.ts"],"names":[],"mappings":"AAEA,OAAO,EAAE,aAAa,EAAE,MAAM,gBAAgB,CAAC;AAC/C,OAAO,EAAE,cAAc,EAAE,MAAM,iBAAiB,CAAC;AACjD,OAAO,EAAqC,UAAU,EAAE,MAAM,wBAAwB,CAAC;AAEvF,OAAO,EAAE,YAAY,EAAE,MAAM,iBAAiB,CAAC;AAC/C,OAAO,EAAE,QAAQ,EAAE,MAAM,qBAAqB,CAAC;AAY/C;;GAEG;AACH,MAAM,CAAC,MAAM,iCAAiC,GAA2B;IACxE,UAAU,EAAE,CAAC,aAAa,EAAE,EAAE,cAAc,EAAE,CAAC;IAC/C,MAAM,EAAE,GAAG;CACX,CAAC;AAEF;;;;;;;;;;;;;;;;;;;;;;;;;;GA0BG;AACH,MAAM,CAAC,MAAM,eAAe,GAAG,CAAC,KAA4B,EAAE,OAAgC,EAAU,EAAE;IACzG,MAAM,EAAE,OAAO,EAAE,GAAG,iBAAiB,EAAE,GAAG,EAAE,GAAG,iCAAiC,EAAE,GAAG,OAAO,EAAE,CAAC;IAE/F,IAAI,QAAQ,GAAG,UAAU,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,iBAAiB,CAAC,CAAC,OAAO,CAAC,WAAW,EAAE,EAAE,CAAC,CAAC;IAEvF,IAAI,OAAO,EAAE,CAAC;QACb,QAAQ,GAAG,YAAY,CAAC,QAAQ,EAAE,OAAO,CAAC,CAAC;IAC5C,CAAC;IAED,OAAO,QAAQ,CAAC;AACjB,CAAC,CAAC"}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"hast-utils.js","sourceRoot":"","sources":["../../../src/utils/hast-utils.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,MAAM,EAAE,SAAS,EAAE,MAAM,kBAAkB,CAAC;AAErD,MAAM,CAAC,MAAM,eAAe,GAAG,CAAC,IAAU,EAAE,EAAE;IAC7C,MAAM,IAAI,GAAG,MAAM,CAAC,MAAM,EAAE,IAAI,CAAC,CAAC;IAClC,IAAI,IAAI,IAAI,OAAO,IAAI,CAAC,UAAU,CAAC,IAAI,KAAK,QAAQ,EAAE,CAAC;QACtD,OAAO,IAAI,CAAC,UAAU,CAAC,IAAI,CAAC;IAC7B,CAAC;IACD,IAAI,IAAI,CAAC,IAAI,KAAK,SAAS,EAAE,CAAC;QAC7B,OAAO;IACR,CAAC;IACD,MAAM,OAAO,GAAG,IAAe,CAAC;IAChC,IAAI,OAAO,CAAC,OAAO,KAAK,MAAM,EAAE,CAAC;QAChC,OAAO;IACR,CAAC;IAED,MAAM,QAAQ,GAAG,OAAO,CAAC,UAAU,CAAC,IAAI,IAAI,OAAO,CAAC,UAAU,CAAC,UAAU,CAAC,CAAC;IAC3E,IAAI,QAAQ,EAAE,CAAC;QACd,OAAO,QAAkB,CAAC;IAC3B,CAAC;IAED,OAAO,SAAS,CAAC;AAClB,CAAC,CAAC;AAEF,MAAM,CAAC,MAAM,cAAc,GAAG,CAAC,GAAW,EAAE,EAAE;IAC7C,MAAM,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC,uBAAuB,CAAC,CAAC;IACjD,IAAI,KAAK,EAAE,CAAC;QACX,OAAO,KAAK,CAAC,CAAC,CAAC,CAAC;IACjB,CAAC;IACD,OAAO,SAAS,CAAC;AAClB,CAAC,CAAC;AAEF,MAAM,CAAC,MAAM,cAAc,GAAG,CAAC,IAAU,EAAsB,EAAE;IAChE,IAAI,IAAI,CAAC,IAAI,KAAK,SAAS,EAAE,CAAC;QAC7B,OAAO,SAAS,CAAC;IAClB,CAAC;IAED,MAAM,iBAAiB,GAAG,CAAC,QAAQ,EAAE,aAAa,CAAC,CAAC;IACpD,MAAM,QAAQ,GAAG,SAAS,CAAC,MAAM,EAAE,IAAI,CAAC,CAAC;IAEzC,KAAK,MAAM,IAAI,IAAI,QAAQ,EAAE,CAAC;QAC7B,MAAM,QAAQ,GAAG,IAAI,CAAC,UAAU,CAAC,QAAQ,IAAI,IAAI,CAAC,UAAU,CAAC,IAAI,CAAC;QAClE,IAAI,OAAO,QAAQ,KAAK,QAAQ,IAAI,iBAAiB,CAAC,QAAQ,CAAC,QAAQ,CAAC,EAAE,CAAC;YAC1E,OAAO,OAAO,IAAI,CAAC,UAAU,CAAC,OAAO,KAAK,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAC,UAAU,CAAC,OAAO,CAAC,CAAC,CAAC,SAAS,CAAC;QAC1F,CAAC;IACF,CAAC;IAED,OAAO,SAAS,CAAC;AAClB,CAAC,CAAC"}
|
|
@@ -0,0 +1,46 @@
|
|
|
1
|
+
// src/utils/hast-utils.ts
|
|
2
|
+
import { select, selectAll } from "hast-util-select";
|
|
3
|
+
var getLangFromHast = (node) => {
|
|
4
|
+
const html = select("html", node);
|
|
5
|
+
if (html && typeof html.properties.lang === "string") {
|
|
6
|
+
return html.properties.lang;
|
|
7
|
+
}
|
|
8
|
+
if (node.type !== "element") {
|
|
9
|
+
return;
|
|
10
|
+
}
|
|
11
|
+
const element = node;
|
|
12
|
+
if (element.tagName !== "html") {
|
|
13
|
+
return;
|
|
14
|
+
}
|
|
15
|
+
const langAttr = element.properties.lang || element.properties["xml:lang"];
|
|
16
|
+
if (langAttr) {
|
|
17
|
+
return langAttr;
|
|
18
|
+
}
|
|
19
|
+
return void 0;
|
|
20
|
+
};
|
|
21
|
+
var getLangFromStr = (str) => {
|
|
22
|
+
const match = str.match(/lang=["']([^"']+)["']/);
|
|
23
|
+
if (match) {
|
|
24
|
+
return match[1];
|
|
25
|
+
}
|
|
26
|
+
return void 0;
|
|
27
|
+
};
|
|
28
|
+
var getUrlFromHast = (node) => {
|
|
29
|
+
if (node.type !== "element") {
|
|
30
|
+
return void 0;
|
|
31
|
+
}
|
|
32
|
+
const metaTagAttributes = ["og:url", "twitter:url"];
|
|
33
|
+
const metaTags = selectAll("meta", node);
|
|
34
|
+
for (const meta of metaTags) {
|
|
35
|
+
const property = meta.properties.property || meta.properties.name;
|
|
36
|
+
if (typeof property === "string" && metaTagAttributes.includes(property)) {
|
|
37
|
+
return typeof meta.properties.content === "string" ? meta.properties.content : void 0;
|
|
38
|
+
}
|
|
39
|
+
}
|
|
40
|
+
return void 0;
|
|
41
|
+
};
|
|
42
|
+
export {
|
|
43
|
+
getLangFromHast,
|
|
44
|
+
getLangFromStr,
|
|
45
|
+
getUrlFromHast
|
|
46
|
+
};
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "webforai",
|
|
3
|
-
"version": "1.
|
|
3
|
+
"version": "1.6.0",
|
|
4
4
|
"description": "A library that provides a web interface for AI",
|
|
5
5
|
"author": "inaridiy",
|
|
6
6
|
"license": "MIT",
|
|
@@ -79,17 +79,17 @@
|
|
|
79
79
|
}
|
|
80
80
|
},
|
|
81
81
|
"peerDependencies": {
|
|
82
|
-
"@cloudflare/puppeteer": "
|
|
83
|
-
"playwright": "
|
|
84
|
-
"puppeteer": "
|
|
82
|
+
"@cloudflare/puppeteer": ">=0.0.6",
|
|
83
|
+
"playwright-core": ">=1.4",
|
|
84
|
+
"puppeteer": ">=22"
|
|
85
85
|
},
|
|
86
86
|
"peerDependenciesMeta": {
|
|
87
|
-
"playwright": {
|
|
88
|
-
"optional": true
|
|
89
|
-
},
|
|
90
87
|
"@cloudflare/puppeteer": {
|
|
91
88
|
"optional": true
|
|
92
89
|
},
|
|
90
|
+
"playwright-core": {
|
|
91
|
+
"optional": false
|
|
92
|
+
},
|
|
93
93
|
"puppeteer": {
|
|
94
94
|
"optional": true
|
|
95
95
|
}
|
|
@@ -97,36 +97,37 @@
|
|
|
97
97
|
"dependencies": {
|
|
98
98
|
"@clack/prompts": "^0.7.0",
|
|
99
99
|
"commander": "^12.1.0",
|
|
100
|
-
"
|
|
101
|
-
"hast-util-
|
|
102
|
-
"hast-util-
|
|
103
|
-
"hast-util-to-html": "^9.0.0",
|
|
100
|
+
"hast-util-from-html": "^2.0.3",
|
|
101
|
+
"hast-util-select": "^6.0.2",
|
|
102
|
+
"hast-util-to-html": "^9.0.3",
|
|
104
103
|
"hast-util-to-mdast": "^10.1.0",
|
|
105
104
|
"hast-util-to-string": "^3.0.0",
|
|
106
105
|
"hast-util-to-text": "^4.0.0",
|
|
107
|
-
"mathml-to-latex": "^1.
|
|
106
|
+
"mathml-to-latex": "^1.4.1",
|
|
108
107
|
"mdast-util-gfm": "^3.0.0",
|
|
109
108
|
"mdast-util-math": "^3.0.0",
|
|
110
109
|
"mdast-util-to-markdown": "^2.1.0",
|
|
111
110
|
"picocolors": "^1.0.1",
|
|
112
|
-
"puppeteer": "^22.10.0",
|
|
113
111
|
"trim-trailing-lines": "^2.1.0",
|
|
114
|
-
"unist-util-filter": "^5.0.1"
|
|
112
|
+
"unist-util-filter": "^5.0.1",
|
|
113
|
+
"zx": "^8.1.5"
|
|
115
114
|
},
|
|
116
115
|
"devDependencies": {
|
|
117
116
|
"@cloudflare/puppeteer": "^0.0.6",
|
|
118
117
|
"@tsconfig/recommended": "^1.0.3",
|
|
118
|
+
"fastest-levenshtein": "^1.0.16",
|
|
119
119
|
"@types/hast": "^3.0.2",
|
|
120
120
|
"@types/mdast": "^4.0.2",
|
|
121
|
-
"@types/node": "^20.
|
|
121
|
+
"@types/node": "^20.14.10",
|
|
122
122
|
"arg": "^5.0.2",
|
|
123
123
|
"esbuild": "^0.19.11",
|
|
124
124
|
"glob": "^10.3.10",
|
|
125
125
|
"ncp": "^2.0.0",
|
|
126
126
|
"np": "^9.2.0",
|
|
127
|
-
"playwright": "^1.40.1",
|
|
127
|
+
"playwright-core": "^1.40.1",
|
|
128
|
+
"puppeteer": "^23.2.2",
|
|
128
129
|
"rimraf": "^5.0.5",
|
|
129
|
-
"tsx": "^4.
|
|
130
|
+
"tsx": "^4.19.1",
|
|
130
131
|
"typescript": "^5.4.5"
|
|
131
132
|
},
|
|
132
133
|
"scripts": {
|
package/dist/cjs/bin.js
DELETED
|
@@ -1,254 +0,0 @@
|
|
|
1
|
-
#!/usr/bin/env node
|
|
2
|
-
"use strict";
|
|
3
|
-
var __create = Object.create;
|
|
4
|
-
var __defProp = Object.defineProperty;
|
|
5
|
-
var __getOwnPropDesc = Object.getOwnPropertyDescriptor;
|
|
6
|
-
var __getOwnPropNames = Object.getOwnPropertyNames;
|
|
7
|
-
var __getProtoOf = Object.getPrototypeOf;
|
|
8
|
-
var __hasOwnProp = Object.prototype.hasOwnProperty;
|
|
9
|
-
var __copyProps = (to, from, except, desc) => {
|
|
10
|
-
if (from && typeof from === "object" || typeof from === "function") {
|
|
11
|
-
for (let key of __getOwnPropNames(from))
|
|
12
|
-
if (!__hasOwnProp.call(to, key) && key !== except)
|
|
13
|
-
__defProp(to, key, { get: () => from[key], enumerable: !(desc = __getOwnPropDesc(from, key)) || desc.enumerable });
|
|
14
|
-
}
|
|
15
|
-
return to;
|
|
16
|
-
};
|
|
17
|
-
var __toESM = (mod, isNodeMode, target) => (target = mod != null ? __create(__getProtoOf(mod)) : {}, __copyProps(
|
|
18
|
-
// If the importer is in node compatibility mode or this is not an ESM
|
|
19
|
-
// file that has been converted to a CommonJS file using a Babel-
|
|
20
|
-
// compatible transform (i.e. "__esModule" has not been set), then set
|
|
21
|
-
// "default" to the CommonJS "module.exports" for node compatibility.
|
|
22
|
-
isNodeMode || !mod || !mod.__esModule ? __defProp(target, "default", { value: mod, enumerable: true }) : target,
|
|
23
|
-
mod
|
|
24
|
-
));
|
|
25
|
-
var import_node_fs = __toESM(require("node:fs"), 1);
|
|
26
|
-
var import_node_path = __toESM(require("node:path"), 1);
|
|
27
|
-
var import_prompts = require("@clack/prompts");
|
|
28
|
-
var import_commander = require("commander");
|
|
29
|
-
var import_picocolors = __toESM(require("picocolors"), 1);
|
|
30
|
-
var import_package = __toESM(require("../package.json"), 1);
|
|
31
|
-
var import_html_to_markdown = require("./html-to-markdown");
|
|
32
|
-
var import_playwright = require("./loaders/playwright");
|
|
33
|
-
var import_puppeteer = require("./loaders/puppeteer");
|
|
34
|
-
var import_bin_utils = require("./utils/bin-utils");
|
|
35
|
-
const MODES = ["default", "ai"];
|
|
36
|
-
const LOADERS = ["fetch", "playwright", "puppeteer"];
|
|
37
|
-
const loadersHint = {
|
|
38
|
-
fetch: "Fetch HTML content from the given URL",
|
|
39
|
-
playwright: "Retrieve HTML content after rendering using Playwright; Playwright must be installed in advance.",
|
|
40
|
-
puppeteer: "Retrieve HTML content after rendering using Puppeteer; Puppeteer must be installed in advance."
|
|
41
|
-
};
|
|
42
|
-
const aiModeOptions = { linkAsText: true, tableAsText: true, hideImage: true };
|
|
43
|
-
const readabilityModeOptions = { linkAsText: false, tableAsText: false, hideImage: false };
|
|
44
|
-
import_commander.program.name("webforai").description("CLI tool for ultra-precise HTML to Markdown conversion").version(import_package.default.version, "-v, --version", "output the current version").argument("[source]", "URL or path to process").argument("[outputPath]", "Path to output file or directory").option("--mode <mode>", `Processing mode (${MODES.join(", ")})`, MODES[0]).option("--loader <loader>", `Loader to use (${LOADERS.join(", ")})`, LOADERS[0]).option("--baseUrl <baseUrl>", "Base URL for relative links").option("-o, --stdout", "output to stdout").option("-d, --debug", "output extra debugging information").on("option:debug", () => {
|
|
45
|
-
process.env.DEBUG = "true";
|
|
46
|
-
}).action(
|
|
47
|
-
async (source, outputPath, options) => {
|
|
48
|
-
let finalSource;
|
|
49
|
-
let sourceIsUrl;
|
|
50
|
-
let finalOutputPath;
|
|
51
|
-
let finalMode;
|
|
52
|
-
let finalLoader = "fetch";
|
|
53
|
-
const isOutputFile = !options.stdout;
|
|
54
|
-
if (process.env.DEBUG) {
|
|
55
|
-
console.info("Arguments and options:");
|
|
56
|
-
console.table({
|
|
57
|
-
source,
|
|
58
|
-
outputPath,
|
|
59
|
-
...options
|
|
60
|
-
});
|
|
61
|
-
console.info("isOutputFile:", isOutputFile);
|
|
62
|
-
}
|
|
63
|
-
try {
|
|
64
|
-
finalSource = await (0, import_prompts.text)({
|
|
65
|
-
message: "Enter the URL or html path to be converted to markdown:",
|
|
66
|
-
placeholder: "https://example.com",
|
|
67
|
-
initialValue: source,
|
|
68
|
-
validate: (value) => {
|
|
69
|
-
if (value.trim() === "") {
|
|
70
|
-
return "Source is required";
|
|
71
|
-
}
|
|
72
|
-
if (!(0, import_bin_utils.isUrl)(value)) {
|
|
73
|
-
if (!import_node_fs.default.existsSync(value)) {
|
|
74
|
-
return "It appears that you are specifying a local file, but the file cannot be found. hint: when specifying a url, start with http or https.";
|
|
75
|
-
}
|
|
76
|
-
if (import_node_fs.default.statSync(value).isDirectory()) {
|
|
77
|
-
return "You are specifying a local file, but you cannot specify a directory. hint: when specifying a url, start with http or https.";
|
|
78
|
-
}
|
|
79
|
-
}
|
|
80
|
-
}
|
|
81
|
-
});
|
|
82
|
-
if ((0, import_prompts.isCancel)(finalSource)) {
|
|
83
|
-
(0, import_prompts.cancel)("Canceled.");
|
|
84
|
-
return;
|
|
85
|
-
}
|
|
86
|
-
sourceIsUrl = (0, import_bin_utils.isUrl)(finalSource);
|
|
87
|
-
if (sourceIsUrl) {
|
|
88
|
-
finalLoader = await (0, import_prompts.select)({
|
|
89
|
-
message: "Select loader:",
|
|
90
|
-
options: LOADERS.map((mode) => ({ value: mode, label: mode, hint: loadersHint[mode] || "" })),
|
|
91
|
-
initialValue: options.loader && LOADERS.includes(options.loader) ? options.loader : "fetch"
|
|
92
|
-
});
|
|
93
|
-
if ((0, import_prompts.isCancel)(finalLoader)) {
|
|
94
|
-
(0, import_prompts.cancel)("Canceled.");
|
|
95
|
-
return;
|
|
96
|
-
}
|
|
97
|
-
}
|
|
98
|
-
if (isOutputFile) {
|
|
99
|
-
finalOutputPath = await (0, import_prompts.text)({
|
|
100
|
-
message: "Enter the output file path:",
|
|
101
|
-
placeholder: "output.md",
|
|
102
|
-
initialValue: outputPath ? outputPath : sourceIsUrl ? `${(0, import_bin_utils.urlToFilename)(finalSource)}.md` : (0, import_bin_utils.changeFileExtension)(finalSource, "md"),
|
|
103
|
-
validate: (value) => {
|
|
104
|
-
if (value.trim() === "") {
|
|
105
|
-
return "Output path is required";
|
|
106
|
-
}
|
|
107
|
-
if (import_node_fs.default.existsSync(value) && import_node_fs.default.statSync(value).isDirectory()) {
|
|
108
|
-
return "No directory can be specified.";
|
|
109
|
-
}
|
|
110
|
-
if (!import_node_fs.default.existsSync(value) && value.endsWith("/")) {
|
|
111
|
-
return "No directory can be specified.";
|
|
112
|
-
}
|
|
113
|
-
}
|
|
114
|
-
});
|
|
115
|
-
if ((0, import_prompts.isCancel)(finalOutputPath)) {
|
|
116
|
-
(0, import_prompts.cancel)("Canceled.");
|
|
117
|
-
return;
|
|
118
|
-
}
|
|
119
|
-
if (import_node_fs.default.existsSync(finalOutputPath)) {
|
|
120
|
-
const isOutputFileOverwrite = await (0, import_prompts.confirm)({
|
|
121
|
-
message: "The file already exists. Overwrite?",
|
|
122
|
-
initialValue: false
|
|
123
|
-
});
|
|
124
|
-
if ((0, import_prompts.isCancel)(isOutputFileOverwrite)) {
|
|
125
|
-
(0, import_prompts.cancel)("Canceled.");
|
|
126
|
-
return;
|
|
127
|
-
}
|
|
128
|
-
if (isOutputFileOverwrite === false) {
|
|
129
|
-
finalOutputPath = await (0, import_prompts.text)({
|
|
130
|
-
message: "Enter the output file path:",
|
|
131
|
-
placeholder: "output.md",
|
|
132
|
-
initialValue: (0, import_bin_utils.getNextAvailableFilePath)(finalOutputPath),
|
|
133
|
-
validate: (value) => {
|
|
134
|
-
if (value.trim() === "") {
|
|
135
|
-
return "Output path is required";
|
|
136
|
-
}
|
|
137
|
-
if (import_node_fs.default.existsSync(value)) {
|
|
138
|
-
return "The file already exists";
|
|
139
|
-
}
|
|
140
|
-
if (!import_node_fs.default.existsSync(value) && value.endsWith("/")) {
|
|
141
|
-
return "No directory can be specified.";
|
|
142
|
-
}
|
|
143
|
-
}
|
|
144
|
-
});
|
|
145
|
-
if ((0, import_prompts.isCancel)(finalOutputPath)) {
|
|
146
|
-
(0, import_prompts.cancel)("Canceled.");
|
|
147
|
-
return;
|
|
148
|
-
}
|
|
149
|
-
}
|
|
150
|
-
}
|
|
151
|
-
} else {
|
|
152
|
-
finalOutputPath = "";
|
|
153
|
-
}
|
|
154
|
-
finalMode = await (0, import_prompts.select)({
|
|
155
|
-
message: "Select processing mode:",
|
|
156
|
-
options: MODES.map((mode) => ({ value: mode, label: mode })),
|
|
157
|
-
initialValue: options.mode && MODES.includes(options.mode) ? options.mode : "default"
|
|
158
|
-
});
|
|
159
|
-
if ((0, import_prompts.isCancel)(finalMode)) {
|
|
160
|
-
(0, import_prompts.cancel)("Canceled.");
|
|
161
|
-
return;
|
|
162
|
-
}
|
|
163
|
-
} catch (e) {
|
|
164
|
-
const error = e;
|
|
165
|
-
console.error(import_picocolors.default.red("An error occurred while processing the given value:"), error.message);
|
|
166
|
-
if (process.env.DEBUG) {
|
|
167
|
-
console.error(error.stack);
|
|
168
|
-
}
|
|
169
|
-
process.exit(1);
|
|
170
|
-
}
|
|
171
|
-
if (process.env.DEBUG) {
|
|
172
|
-
console.info("Processing with the following options:");
|
|
173
|
-
console.table({ finalSource, sourceIsUrl, isOutputFile, finalOutputPath, finalMode, finalLoader });
|
|
174
|
-
}
|
|
175
|
-
let content;
|
|
176
|
-
const loadSpinner = (0, import_prompts.spinner)();
|
|
177
|
-
try {
|
|
178
|
-
if (sourceIsUrl) {
|
|
179
|
-
loadSpinner.start("Downloading content.");
|
|
180
|
-
if (finalLoader === "playwright") {
|
|
181
|
-
content = await (0, import_playwright.loadHtml)(finalSource);
|
|
182
|
-
} else if (finalLoader === "puppeteer") {
|
|
183
|
-
content = await (0, import_puppeteer.loadHtml)(finalSource);
|
|
184
|
-
} else {
|
|
185
|
-
const response = await fetch(finalSource);
|
|
186
|
-
content = await response.text();
|
|
187
|
-
}
|
|
188
|
-
loadSpinner.stop("Content download is complete!");
|
|
189
|
-
} else {
|
|
190
|
-
loadSpinner.start("Retrieving content.");
|
|
191
|
-
if (!import_node_fs.default.existsSync(finalSource)) {
|
|
192
|
-
console.error(import_picocolors.default.red("The file does not exist"));
|
|
193
|
-
process.exit(1);
|
|
194
|
-
}
|
|
195
|
-
const stats = import_node_fs.default.statSync(finalSource);
|
|
196
|
-
if (!stats.isFile()) {
|
|
197
|
-
console.error(import_picocolors.default.red("The source must be a file"));
|
|
198
|
-
process.exit(1);
|
|
199
|
-
}
|
|
200
|
-
content = import_node_fs.default.readFileSync(finalSource, "utf-8");
|
|
201
|
-
loadSpinner.stop("Content retrieval is complete!");
|
|
202
|
-
}
|
|
203
|
-
} catch (e) {
|
|
204
|
-
const error = e;
|
|
205
|
-
loadSpinner.stop(import_picocolors.default.red("An error occurred while retrieving content"));
|
|
206
|
-
console.error(error.message);
|
|
207
|
-
if (process.env.DEBUG) {
|
|
208
|
-
console.error(error.stack);
|
|
209
|
-
}
|
|
210
|
-
process.exit(1);
|
|
211
|
-
}
|
|
212
|
-
if (process.env.DEBUG) {
|
|
213
|
-
console.info("Content:");
|
|
214
|
-
console.info(content);
|
|
215
|
-
}
|
|
216
|
-
const convertSpinner = (0, import_prompts.spinner)();
|
|
217
|
-
try {
|
|
218
|
-
convertSpinner.start("Converting content to markdown.");
|
|
219
|
-
const markdown = (0, import_html_to_markdown.htmlToMarkdown)(content, {
|
|
220
|
-
...sourceIsUrl || !!options.baseUrl && options.baseUrl.trim() !== "" ? { baseUrl: options.baseUrl || finalSource } : {},
|
|
221
|
-
...finalMode === "ai" ? aiModeOptions : readabilityModeOptions
|
|
222
|
-
});
|
|
223
|
-
if (process.env.DEBUG) {
|
|
224
|
-
console.info("Markdown:");
|
|
225
|
-
console.info(markdown);
|
|
226
|
-
}
|
|
227
|
-
if (isOutputFile) {
|
|
228
|
-
const directory = import_node_path.default.dirname(finalOutputPath);
|
|
229
|
-
if (!import_node_fs.default.existsSync(directory)) {
|
|
230
|
-
import_node_fs.default.mkdirSync(directory, { recursive: true });
|
|
231
|
-
}
|
|
232
|
-
import_node_fs.default.writeFileSync(finalOutputPath, markdown);
|
|
233
|
-
convertSpinner.stop("Markdown conversion is complete!");
|
|
234
|
-
(0, import_prompts.outro)(`Converted successfully to markdown! Output saved to: ${import_node_path.default.resolve(finalOutputPath)} \u{1F389}`);
|
|
235
|
-
} else {
|
|
236
|
-
convertSpinner.stop("Markdown conversion is complete!");
|
|
237
|
-
(0, import_prompts.outro)("Converted successfully to markdown\u{1F389}");
|
|
238
|
-
console.log(`
|
|
239
|
-
|
|
240
|
-
${markdown}`);
|
|
241
|
-
}
|
|
242
|
-
} catch (e) {
|
|
243
|
-
const error = e;
|
|
244
|
-
convertSpinner.stop(import_picocolors.default.red("An error occurred while converting and saving the markdown"));
|
|
245
|
-
console.error(error.message);
|
|
246
|
-
if (process.env.DEBUG) {
|
|
247
|
-
console.error(error.stack);
|
|
248
|
-
}
|
|
249
|
-
process.exit(1);
|
|
250
|
-
}
|
|
251
|
-
}
|
|
252
|
-
);
|
|
253
|
-
(0, import_prompts.intro)(import_picocolors.default.inverse(`webforai CLI version ${import_package.default.version}`));
|
|
254
|
-
import_commander.program.parse();
|