var _ = require('lodash'); module.exports = function regionExtractor() { var nullLine = '###'; var nullLinePattern = new RegExp(nullLine + '\n', 'g'); // split out each fragment in {content} into a separate doc // a fragment is a section of text surrounded by // 1) In front: a comment marker followed by '#docregion' followed by an optional region name. For example: // <-- #docregion foo --> for html // or // #docregion foo for js/ts // 2) In back: a comment marker followed by '#enddocregion' // Regions can be nested and any regions not 'closed' are automatically closed at the end of the doc. // empty enddocregion always closes last region started. // enddocregions with names that do no match start region tags get ignored. return function(content, commentPrefixes) { var lines = result = content.split(/\r?\n/); var docStack = []; var docMap = {}; var doc; var regionNames; lines.forEach(function(line, ix) { if (isCommentLine(line, commentPrefixes)) { if (hasRegionTag(line)) { lines[ix] = nullLine; regionNames = getRegionNames(line); regionNames.forEach(function(rn) { doc = docMap[rn]; if (!doc) { // regionName may be '' doc = {regionName: rn, ranges: [ { startIx: ix} ] }; docMap[rn] = doc; } else { // only add a new range if prev range is closed var lastRange = doc.ranges[doc.ranges.length-1]; if (lastRange.endIx) { doc.ranges.push({startIx: ix}); } } docStack.push(doc); }); } else if (hasEndRegionTag(line)) { lines[ix] = nullLine; regionNames = getEndRegionNames(line); regionNames.forEach(function(rn) { // handle endregions with no name specially. // They operate on the last region created. if (rn.length == 0) { if (docStack.length) { // update last item on the stack doc = docStack.pop(); doc.ranges[doc.ranges.length - 1].endIx = ix; } } else { doc = docMap[rn]; // ignore endregion if name is specified but not found. if (doc) { doc.ranges[doc.ranges.length - 1].endIx = ix; // remove doc from stack _.remove(docStack, function (item) { return item.regionName === rn; }); } } }) } } }); var docs = _.values(docMap); docs.forEach(function(doc) { var content; var fragLines = []; doc.ranges.forEach(function (range) { var subLines; if (range.endIx) { subLines = lines.slice(range.startIx + 1, range.endIx); } else { subLines = lines.slice(range.startIx + 1); } fragLines = fragLines.concat(subLines); }); fragLines = trimLeftIndent(fragLines); content = fragLines.join('\n'); // eliminate all #docregion lines content = content.replace(nullLinePattern, ''); if (content.substr(-3) === nullLine) { content = content.substr(0, content.length-3); } doc.content = content; }); return docs; } }; function trimLeftIndent(lines) { var minIx = 100; var ok = lines.every(function(line) { // var ix = line.search(/\S/); var ix = line.search(/[^ ]/); if (ix === 0) return false; if (ix === -1) return true; if (ix > 0) { minIx = Math.min(minIx, ix); } return true; }); if ( (!ok) || minIx === 100) return lines; var result = lines.map(function(line) { if (line.length > minIx) { return line.substr(minIx); } else { // this can happen if line is all blanks and shorter than mixIx return line; } }); return result; } function isCommentLine(line, commentPrefixes) { return commentPrefixes.some(function(prefix) { return line.trim().indexOf(prefix) == 0; }); } function hasRegionTag(line) { return line.indexOf("#docregion") >= 0; } function hasEndRegionTag(line) { return line.indexOf("#enddocregion") >= 0; } function getRegionNames(line) { return extractRegionNames(line, /#docregion\s*(\S.*)/); } function getEndRegionNames(line) { return extractRegionNames(line, /#enddocregion\s*(\S.*)/); } function extractRegionNames(line, rx) { try { var names = line.match(rx)[1]; names = names.replace(/\s*/g,''); // Hack for html regions that look like or */ names = names.replace("-->","").replace('\*\/',""); names = names.split(','); return names; } catch (e) { return ['']; } }