{"version":3,"sources":["../../lib/Tabletojson.ts"],"sourcesContent":["import * as cheerio from 'cheerio';\nimport {CallbackFunction, TableToJsonOptions} from './index';\n\nexport class Tabletojson {\n    static convert(\n        html: string,\n        options: TableToJsonOptions = {\n            useFirstRowForHeadings: false,\n            stripHtmlFromHeadings: true,\n            stripHtmlFromCells: true,\n            stripHtml: null,\n            forceIndexAsNumber: false,\n            countDuplicateHeadings: true,\n            ignoreColumns: null,\n            onlyColumns: null,\n            ignoreHiddenRows: true,\n            headings: null,\n            containsClasses: null,\n            id: null,\n            limitrows: null,\n        },\n    ): any[] {\n        options = Object.assign(\n            {\n                useFirstRowForHeadings: false,\n                stripHtmlFromHeadings: true,\n                stripHtmlFromCells: true,\n                stripHtml: null,\n                forceIndexAsNumber: false,\n                countDuplicateHeadings: true,\n                ignoreColumns: null,\n                onlyColumns: null,\n                ignoreHiddenRows: true,\n                headings: null,\n                containsClasses: null,\n                id: null,\n                limitrows: null,\n            },\n            options,\n        );\n\n        if (options.stripHtml === true) {\n            options.stripHtmlFromHeadings = true;\n            options.stripHtmlFromCells = true;\n        } else if (options.stripHtml === false) {\n            options.stripHtmlFromHeadings = false;\n            options.stripHtmlFromCells = false;\n        }\n\n        const jsonResponse: any[] = [];\n        let suffix;\n\n        const $ = cheerio.load(html);\n\n        let additionalSelectors = options.containsClasses ? `.${options.containsClasses.join('.')}` : '';\n        additionalSelectors = options.id ? `${additionalSelectors}#${options.id}` : '';\n\n        $(`table${additionalSelectors}`).each((_i, table) => {\n            const tableAsJson: any[] = [];\n            const alreadySeen: any = {};\n            // Get column headings\n            // @fixme Doesn't support vertical column headings.\n            // @todo Try to support badly formated tables.\n            const columnHeadings: string[] = [];\n\n            let trs: cheerio.Cheerio = $(table).find('tr');\n\n            if (options.useFirstRowForHeadings) {\n                trs = $(trs[0]);\n            }\n            let headingsCounter: number = 0;\n            // Use headings for objects key evaluation\n            trs.each((_index: number, row: cheerio.Element) => {\n                const cells: cheerio.Cheerio = options.useFirstRowForHeadings\n                    ? $(row).find('td, th')\n                    : $(row).find('th');\n                cells.each((j: number, cell: cheerio.Element) => {\n                    if (options.onlyColumns && !options.onlyColumns.includes(j)) return;\n                    if (options.ignoreColumns && !options.onlyColumns && options.ignoreColumns.includes(j)) return;\n                    let value: string = '';\n\n                    if (options.headings) {\n                        value = options.headings[headingsCounter++];\n                    } else {\n                        const cheerioCell: cheerio.Cheerio = $(cell);\n                        const cheerioCellText: string = cheerioCell.text();\n                        const cheerioCellHtml: string | null = cheerioCell.html();\n\n                        value = options.stripHtmlFromHeadings\n                            ? cheerioCellText.trim()\n                            : cheerioCellHtml\n                            ? cheerioCellHtml.trim()\n                            : '';\n                    }\n\n                    const seen: any = alreadySeen[value];\n                    if (seen && options.countDuplicateHeadings) {\n                        suffix = ++alreadySeen[value];\n                        columnHeadings[j] = value !== '' ? `${value}_${suffix}` : `${j}`;\n                    } else {\n                        alreadySeen[value] = 1;\n                        columnHeadings[j] = value;\n                    }\n                });\n            });\n\n            let rowspans: any[] = [];\n\n            // Fetch each row\n            $(table)\n                .find('tr')\n                .each(function (i, row) {\n                    const rowAsJson: any = {};\n\n                    function setColumn(j: number, content: string) {\n                        if (columnHeadings[j] && !options.forceIndexAsNumber) {\n                            rowAsJson[columnHeadings[j]] = content;\n                        } else {\n                            rowAsJson[j] = content;\n                        }\n                    }\n\n                    // Add content from rowspans\n                    rowspans.forEach((rowspan: any, index: number) => {\n                        if (!rowspan) return;\n\n                        setColumn(index, rowspan.content);\n\n                        rowspan.value--;\n                    });\n                    const nextrowspans: any[] = [...rowspans];\n\n                    const cells: cheerio.Cheerio = options.useFirstRowForHeadings\n                        ? $(row).find('td, th')\n                        : $(row).find('td');\n                    cells.each((j: number, cell: cheerio.Element) => {\n                        // ignoreHiddenRows\n                        if (options.ignoreHiddenRows) {\n                            const style: string | undefined = $(row).attr('style');\n                            if (style) {\n                                const m = style.match(/.*display.*:.*none.*/g);\n                                if (m && m.length > 0) return;\n                            }\n                        }\n\n                        // Apply rowspans offsets\n                        let aux: number = j;\n                        j = 0;\n                        do {\n                            while (rowspans[j]) j++;\n                            while (aux && !rowspans[j]) {\n                                j++;\n                                aux--;\n                            }\n                        } while (aux);\n\n                        if (options.onlyColumns && !options.onlyColumns.includes(j)) return;\n                        if (options.ignoreColumns && !options.onlyColumns && options.ignoreColumns.includes(j)) return;\n\n                        const cheerioCell: cheerio.Cheerio = $(cell);\n                        const cheerioCellText: string = cheerioCell.text();\n                        const cheerioCellHtml: string | null = cheerioCell.html();\n                        const cheerioCellRowspan: string | undefined = cheerioCell.attr('rowspan');\n\n                        const content: string = options.stripHtmlFromCells\n                            ? cheerioCellText.trim()\n                            : cheerioCellHtml\n                            ? cheerioCellHtml.trim()\n                            : '';\n\n                        setColumn(j, content);\n\n                        // Check rowspan\n                        const value: number = cheerioCellRowspan ? parseInt(cheerioCellRowspan, 10) - 1 : 0;\n                        if (value > 0) nextrowspans[j] = {content, value};\n                    });\n\n                    rowspans = nextrowspans;\n                    rowspans.forEach((rowspan: any, index: number) => {\n                        if (rowspan && rowspan.value === 0) rowspans[index] = null;\n                    });\n\n                    // Skip blank rows\n                    if (JSON.stringify(rowAsJson) !== '{}') tableAsJson.push(rowAsJson);\n\n                    if (options.limitrows && i === options.limitrows) {\n                        return false;\n                    }\n                });\n\n            // Add the table to the response\n            const dataContained: boolean = tableAsJson.length > 0;\n            const pushToJsonResult: boolean = Array.isArray(tableAsJson) && dataContained;\n            if (!pushToJsonResult) {\n                return true;\n            }\n            jsonResponse.push(tableAsJson);\n        });\n\n        return jsonResponse;\n    }\n\n    /**\n     * Convert an HTML Page for a given URL\n     * @param url URL to be called\n     * @param callbackFunctionOrOptions {Object} Options for html conversion\n     * @param callbackFunctionOrOptions.useFirstRowForHeadings Use the first row as header [default=false]\n     * @param callbackFunctionOrOptions.stripHtmlFromHeadings Strip all HTML from headings [default=true]\n     * @param callbackFunctionOrOptions.stripHtmlFromCells Strip HTML from cells [default=true]\n     * @param callbackFunctionOrOptions.stripHtml Strip off HTML [default=null] if set true stripHtmlFromHeadings and stripHtmlFromCells will also be true\n     * @param callbackFunctionOrOptions.forceIndexAsNumber Force the index to be used as number [default=false]\n     * @param callbackFunctionOrOptions.countDuplicateHeadings If given a _<NUMBER> will be added to the duplicate key [default=false]\n     * @param callbackFunctionOrOptions.ignoreColumns {Array} Array of column indices to ignored [default=null]\n     * @param callbackFunctionOrOptions.onlyColumns {Array} Array of column indices to be used. Overrides ignoreColumn [default=null]\n     * @param callbackFunctionOrOptions.ignoreHiddenRows Ignoring hidden rows [default=true]\n     * @param callbackFunctionOrOptions.headings {Array} Array of Strings to be used as headings [default=null]\n     * @param callbackFunctionOrOptions.headings {Array} Array of classes to find a specific table [default=null]\n     * @param callbackFunctionOrOptions.limitrows {Integer} Integer that limits the result of all rows to a given amount of data [default=null]\n     * @param callbackFunctionOrOptions.request Options to be passed to request object\n     * @param callbackFunction Callback function to be called when the conversion finished\n     * @return {Promise<*>} Promise containing the result\n     */\n    static async convertUrl(\n        url: string,\n        callbackFunctionOrOptions?: TableToJsonOptions | CallbackFunction,\n        callbackFunction?: CallbackFunction,\n    ): Promise<any> {\n        let options: TableToJsonOptions;\n        let callback = null;\n        let fetchOptions: RequestInit;\n\n        if (\n            callbackFunction &&\n            typeof callbackFunction === 'function' &&\n            typeof callbackFunctionOrOptions === 'object'\n        ) {\n            // If both options and callback passed\n            options = callbackFunctionOrOptions;\n            // If you need to pass in options for request (proxy)\n            // add them to callbackFunctionOrOptions.request\n            fetchOptions = options.fetchOptions || {};\n            callback = callbackFunction;\n\n            // Use a callback (if passed)\n            const result = await fetch(url, fetchOptions);\n            const resultMimetype = result.headers.get('content-type');\n            if (resultMimetype && !resultMimetype.includes('text/')) {\n                throw new Error('Tabletojson can just handle text/** mimetypes');\n            }\n            return callback.call(this, Tabletojson.convert(await result.text(), options));\n        } else if (typeof callbackFunctionOrOptions === 'function') {\n            // If only callback passed, invoke with no options\n            callback = callbackFunctionOrOptions;\n\n            // Use a callback (if passed)\n            const result = await fetch(url);\n            const resultMimetype = result.headers.get('content-type');\n            if (resultMimetype && !resultMimetype.includes('text/')) {\n                throw new Error('Tabletojson can just handle text/** mimetypes');\n            }\n            return callback.call(this, Tabletojson.convert(await result.text()));\n        } else {\n            // If neither argument is callback, return a promise\n            options = callbackFunctionOrOptions || {};\n            // If you need to pass in options for request (proxy)\n            // add them to callbackFunctionOrOptions.request\n            fetchOptions = options.fetchOptions || {};\n            const result = await fetch(url);\n            const resultMimetype = result.headers.get('content-type');\n            if (resultMimetype && !resultMimetype.includes('text/')) {\n                throw new Error('Tabletojson can just handle text/** mimetypes');\n            }\n            return Tabletojson.convert(await result.text(), options);\n        }\n    }\n}\n"],"names":["cheerio","Tabletojson","convert","html","options","useFirstRowForHeadings","stripHtmlFromHeadings","stripHtmlFromCells","stripHtml","forceIndexAsNumber","countDuplicateHeadings","ignoreColumns","onlyColumns","ignoreHiddenRows","headings","containsClasses","id","limitrows","Object","assign","jsonResponse","suffix","$","load","additionalSelectors","join","each","_i","table","tableAsJson","alreadySeen","columnHeadings","trs","find","headingsCounter","_index","row","cells","j","cell","includes","value","cheerioCell","cheerioCellText","text","cheerioCellHtml","trim","seen","rowspans","i","rowAsJson","setColumn","content","forEach","rowspan","index","nextrowspans","style","attr","m","match","length","aux","cheerioCellRowspan","parseInt","JSON","stringify","push","dataContained","pushToJsonResult","Array","isArray","convertUrl","url","callbackFunctionOrOptions","callbackFunction","callback","fetchOptions","result","fetch","resultMimetype","headers","get","Error","call"],"mappings":"AAAA,YAAYA,aAAa,UAAU;AAGnC,OAAO,MAAMC;IACT,OAAOC,QACHC,IAAY,EACZC,UAA8B;QAC1BC,wBAAwB;QACxBC,uBAAuB;QACvBC,oBAAoB;QACpBC,WAAW;QACXC,oBAAoB;QACpBC,wBAAwB;QACxBC,eAAe;QACfC,aAAa;QACbC,kBAAkB;QAClBC,UAAU;QACVC,iBAAiB;QACjBC,IAAI;QACJC,WAAW;IACf,CAAC,EACI;QACLb,UAAUc,OAAOC,MAAM,CACnB;YACId,wBAAwB;YACxBC,uBAAuB;YACvBC,oBAAoB;YACpBC,WAAW;YACXC,oBAAoB;YACpBC,wBAAwB;YACxBC,eAAe;YACfC,aAAa;YACbC,kBAAkB;YAClBC,UAAU;YACVC,iBAAiB;YACjBC,IAAI;YACJC,WAAW;QACf,GACAb;QAGJ,IAAIA,QAAQI,SAAS,KAAK,MAAM;YAC5BJ,QAAQE,qBAAqB,GAAG;YAChCF,QAAQG,kBAAkB,GAAG;QACjC,OAAO,IAAIH,QAAQI,SAAS,KAAK,OAAO;YACpCJ,QAAQE,qBAAqB,GAAG;YAChCF,QAAQG,kBAAkB,GAAG;QACjC;QAEA,MAAMa,eAAsB,EAAE;QAC9B,IAAIC;QAEJ,MAAMC,IAAItB,QAAQuB,IAAI,CAACpB;QAEvB,IAAIqB,sBAAsBpB,QAAQW,eAAe,GAAG,CAAC,CAAC,EAAEX,QAAQW,eAAe,CAACU,IAAI,CAAC,KAAK,CAAC,GAAG;QAC9FD,sBAAsBpB,QAAQY,EAAE,GAAG,CAAC,EAAEQ,oBAAoB,CAAC,EAAEpB,QAAQY,EAAE,CAAC,CAAC,GAAG;QAE5EM,EAAE,CAAC,KAAK,EAAEE,oBAAoB,CAAC,EAAEE,IAAI,CAAC,CAACC,IAAIC;YACvC,MAAMC,cAAqB,EAAE;YAC7B,MAAMC,cAAmB,CAAC;YAC1B,sBAAsB;YACtB,mDAAmD;YACnD,8CAA8C;YAC9C,MAAMC,iBAA2B,EAAE;YAEnC,IAAIC,MAAuBV,EAAEM,OAAOK,IAAI,CAAC;YAEzC,IAAI7B,QAAQC,sBAAsB,EAAE;gBAChC2B,MAAMV,EAAEU,GAAG,CAAC,EAAE;YAClB;YACA,IAAIE,kBAA0B;YAC9B,0CAA0C;YAC1CF,IAAIN,IAAI,CAAC,CAACS,QAAgBC;gBACtB,MAAMC,QAAyBjC,QAAQC,sBAAsB,GACvDiB,EAAEc,KAAKH,IAAI,CAAC,YACZX,EAAEc,KAAKH,IAAI,CAAC;gBAClBI,MAAMX,IAAI,CAAC,CAACY,GAAWC;oBACnB,IAAInC,QAAQQ,WAAW,IAAI,CAACR,QAAQQ,WAAW,CAAC4B,QAAQ,CAACF,IAAI;oBAC7D,IAAIlC,QAAQO,aAAa,IAAI,CAACP,QAAQQ,WAAW,IAAIR,QAAQO,aAAa,CAAC6B,QAAQ,CAACF,IAAI;oBACxF,IAAIG,QAAgB;oBAEpB,IAAIrC,QAAQU,QAAQ,EAAE;wBAClB2B,QAAQrC,QAAQU,QAAQ,CAACoB,kBAAkB;oBAC/C,OAAO;wBACH,MAAMQ,cAA+BpB,EAAEiB;wBACvC,MAAMI,kBAA0BD,YAAYE,IAAI;wBAChD,MAAMC,kBAAiCH,YAAYvC,IAAI;wBAEvDsC,QAAQrC,QAAQE,qBAAqB,GAC/BqC,gBAAgBG,IAAI,KACpBD,kBACAA,gBAAgBC,IAAI,KACpB;oBACV;oBAEA,MAAMC,OAAYjB,WAAW,CAACW,MAAM;oBACpC,IAAIM,QAAQ3C,QAAQM,sBAAsB,EAAE;wBACxCW,SAAS,EAAES,WAAW,CAACW,MAAM;wBAC7BV,cAAc,CAACO,EAAE,GAAGG,UAAU,KAAK,CAAC,EAAEA,MAAM,CAAC,EAAEpB,OAAO,CAAC,GAAG,CAAC,EAAEiB,EAAE,CAAC;oBACpE,OAAO;wBACHR,WAAW,CAACW,MAAM,GAAG;wBACrBV,cAAc,CAACO,EAAE,GAAGG;oBACxB;gBACJ;YACJ;YAEA,IAAIO,WAAkB,EAAE;YAExB,iBAAiB;YACjB1B,EAAEM,OACGK,IAAI,CAAC,MACLP,IAAI,CAAC,SAAUuB,CAAC,EAAEb,GAAG;gBAClB,MAAMc,YAAiB,CAAC;gBAExB,SAASC,UAAUb,CAAS,EAAEc,OAAe;oBACzC,IAAIrB,cAAc,CAACO,EAAE,IAAI,CAAClC,QAAQK,kBAAkB,EAAE;wBAClDyC,SAAS,CAACnB,cAAc,CAACO,EAAE,CAAC,GAAGc;oBACnC,OAAO;wBACHF,SAAS,CAACZ,EAAE,GAAGc;oBACnB;gBACJ;gBAEA,4BAA4B;gBAC5BJ,SAASK,OAAO,CAAC,CAACC,SAAcC;oBAC5B,IAAI,CAACD,SAAS;oBAEdH,UAAUI,OAAOD,QAAQF,OAAO;oBAEhCE,QAAQb,KAAK;gBACjB;gBACA,MAAMe,eAAsB;uBAAIR;iBAAS;gBAEzC,MAAMX,QAAyBjC,QAAQC,sBAAsB,GACvDiB,EAAEc,KAAKH,IAAI,CAAC,YACZX,EAAEc,KAAKH,IAAI,CAAC;gBAClBI,MAAMX,IAAI,CAAC,CAACY,GAAWC;oBACnB,mBAAmB;oBACnB,IAAInC,QAAQS,gBAAgB,EAAE;wBAC1B,MAAM4C,QAA4BnC,EAAEc,KAAKsB,IAAI,CAAC;wBAC9C,IAAID,OAAO;4BACP,MAAME,IAAIF,MAAMG,KAAK,CAAC;4BACtB,IAAID,KAAKA,EAAEE,MAAM,GAAG,GAAG;wBAC3B;oBACJ;oBAEA,yBAAyB;oBACzB,IAAIC,MAAcxB;oBAClBA,IAAI;oBACJ,GAAG;wBACC,MAAOU,QAAQ,CAACV,EAAE,CAAEA;wBACpB,MAAOwB,OAAO,CAACd,QAAQ,CAACV,EAAE,CAAE;4BACxBA;4BACAwB;wBACJ;oBACJ,QAASA,IAAK;oBAEd,IAAI1D,QAAQQ,WAAW,IAAI,CAACR,QAAQQ,WAAW,CAAC4B,QAAQ,CAACF,IAAI;oBAC7D,IAAIlC,QAAQO,aAAa,IAAI,CAACP,QAAQQ,WAAW,IAAIR,QAAQO,aAAa,CAAC6B,QAAQ,CAACF,IAAI;oBAExF,MAAMI,cAA+BpB,EAAEiB;oBACvC,MAAMI,kBAA0BD,YAAYE,IAAI;oBAChD,MAAMC,kBAAiCH,YAAYvC,IAAI;oBACvD,MAAM4D,qBAAyCrB,YAAYgB,IAAI,CAAC;oBAEhE,MAAMN,UAAkBhD,QAAQG,kBAAkB,GAC5CoC,gBAAgBG,IAAI,KACpBD,kBACAA,gBAAgBC,IAAI,KACpB;oBAENK,UAAUb,GAAGc;oBAEb,gBAAgB;oBAChB,MAAMX,QAAgBsB,qBAAqBC,SAASD,oBAAoB,MAAM,IAAI;oBAClF,IAAItB,QAAQ,GAAGe,YAAY,CAAClB,EAAE,GAAG;wBAACc;wBAASX;oBAAK;gBACpD;gBAEAO,WAAWQ;gBACXR,SAASK,OAAO,CAAC,CAACC,SAAcC;oBAC5B,IAAID,WAAWA,QAAQb,KAAK,KAAK,GAAGO,QAAQ,CAACO,MAAM,GAAG;gBAC1D;gBAEA,kBAAkB;gBAClB,IAAIU,KAAKC,SAAS,CAAChB,eAAe,MAAMrB,YAAYsC,IAAI,CAACjB;gBAEzD,IAAI9C,QAAQa,SAAS,IAAIgC,MAAM7C,QAAQa,SAAS,EAAE;oBAC9C,OAAO;gBACX;YACJ;YAEJ,gCAAgC;YAChC,MAAMmD,gBAAyBvC,YAAYgC,MAAM,GAAG;YACpD,MAAMQ,mBAA4BC,MAAMC,OAAO,CAAC1C,gBAAgBuC;YAChE,IAAI,CAACC,kBAAkB;gBACnB,OAAO;YACX;YACAjD,aAAa+C,IAAI,CAACtC;QACtB;QAEA,OAAOT;IACX;IAEA;;;;;;;;;;;;;;;;;;;KAmBC,GACD,aAAaoD,WACTC,GAAW,EACXC,yBAAiE,EACjEC,gBAAmC,EACvB;QACZ,IAAIvE;QACJ,IAAIwE,WAAW;QACf,IAAIC;QAEJ,IACIF,oBACA,OAAOA,qBAAqB,cAC5B,OAAOD,8BAA8B,UACvC;YACE,sCAAsC;YACtCtE,UAAUsE;YACV,qDAAqD;YACrD,gDAAgD;YAChDG,eAAezE,QAAQyE,YAAY,IAAI,CAAC;YACxCD,WAAWD;YAEX,6BAA6B;YAC7B,MAAMG,SAAS,MAAMC,MAAMN,KAAKI;YAChC,MAAMG,iBAAiBF,OAAOG,OAAO,CAACC,GAAG,CAAC;YAC1C,IAAIF,kBAAkB,CAACA,eAAexC,QAAQ,CAAC,UAAU;gBACrD,MAAM,IAAI2C,MAAM;YACpB;YACA,OAAOP,SAASQ,IAAI,CAAC,IAAI,EAAEnF,YAAYC,OAAO,CAAC,MAAM4E,OAAOlC,IAAI,IAAIxC;QACxE,OAAO,IAAI,OAAOsE,8BAA8B,YAAY;YACxD,kDAAkD;YAClDE,WAAWF;YAEX,6BAA6B;YAC7B,MAAMI,SAAS,MAAMC,MAAMN;YAC3B,MAAMO,iBAAiBF,OAAOG,OAAO,CAACC,GAAG,CAAC;YAC1C,IAAIF,kBAAkB,CAACA,eAAexC,QAAQ,CAAC,UAAU;gBACrD,MAAM,IAAI2C,MAAM;YACpB;YACA,OAAOP,SAASQ,IAAI,CAAC,IAAI,EAAEnF,YAAYC,OAAO,CAAC,MAAM4E,OAAOlC,IAAI;QACpE,OAAO;YACH,oDAAoD;YACpDxC,UAAUsE,6BAA6B,CAAC;YACxC,qDAAqD;YACrD,gDAAgD;YAChDG,eAAezE,QAAQyE,YAAY,IAAI,CAAC;YACxC,MAAMC,SAAS,MAAMC,MAAMN;YAC3B,MAAMO,iBAAiBF,OAAOG,OAAO,CAACC,GAAG,CAAC;YAC1C,IAAIF,kBAAkB,CAACA,eAAexC,QAAQ,CAAC,UAAU;gBACrD,MAAM,IAAI2C,MAAM;YACpB;YACA,OAAOlF,YAAYC,OAAO,CAAC,MAAM4E,OAAOlC,IAAI,IAAIxC;QACpD;IACJ;AACJ"}