aaron.axvig/firewallissues
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Parser refactor

Browse Source
This commit is contained in:
aaron.axvig
2026-03-20 08:51:46 -05:00
parent 257a75e1ed
commit 72574b2fca
2 changed files with 64 additions and 60 deletions
Download Patch File Download Diff File Expand all files Collapse all files
test/process-parser.test.mjs
+1 -2
View File
@@ -54,11 +54,10 @@ fixtures.forEach(fixtureId => {
metadata[key.trim()] = valueParts.join(':').trim(); metadata[key.trim()] = valueParts.join(':').trim();
}); });
const parsedIssues = parseIssuesFromHtmlTable(inputHtml);
// Extract type from folder name (e.g., "PAA-25.6.2-known" → "Known") // Extract type from folder name (e.g., "PAA-25.6.2-known" → "Known")
const typeFromFolder = fixtureId.split('-').pop().toLowerCase(); const typeFromFolder = fixtureId.split('-').pop().toLowerCase();
const capitalizedType = typeFromFolder.charAt(0).toUpperCase() + typeFromFolder.slice(1); const capitalizedType = typeFromFolder.charAt(0).toUpperCase() + typeFromFolder.slice(1);
const parsedIssues = parseIssuesFromHtmlTable(inputHtml, { type: capitalizedType });
const markdown = buildIssueMarkdownDocument({ const markdown = buildIssueMarkdownDocument({
type: capitalizedType, type: capitalizedType,
web/js/process.js
+63 -58
View File
@@ -14,7 +14,7 @@ if (typeof document !== 'undefined') {
let activeDownloadUrl = null; let activeDownloadUrl = null;
const PROCESS_FORM_STATE_KEY = 'bugmedley.process.formState.v1'; const PROCESS_FORM_STATE_KEY = 'bugmedley.process.formState.v1';
const ISSUE_ID_PATTERN = /\b((?:[A-Z]{2,6}|WF500)-\d{4,8})\b/i; const ISSUE_ID_PATTERN = /((?:[A-Z]{2,6}|WF500)-\d{4,8})/i;
const BLOCK_CONTAINER_TAGS = new Set(['ARTICLE', 'ASIDE', 'BLOCKQUOTE', 'DIV', 'P', 'PRE', 'SECTION']); const BLOCK_CONTAINER_TAGS = new Set(['ARTICLE', 'ASIDE', 'BLOCKQUOTE', 'DIV', 'P', 'PRE', 'SECTION']);
const RESOLVED_LINE_PATTERN = /^(?:resolved|fixed)\s+in\b/i; const RESOLVED_LINE_PATTERN = /^(?:resolved|fixed)\s+in\b/i;
const TEXT_NODE = 3; const TEXT_NODE = 3;
@@ -69,7 +69,7 @@ function generateJSON() {
return; return;
} }
const parsedIssues = parseIssuesFromHtmlTable(inputText); const parsedIssues = parseIssuesFromHtmlTable(inputText, { type: issueType });
if (parsedIssues.length === 0) { if (parsedIssues.length === 0) {
document.getElementById('markdownOutput').value = ''; document.getElementById('markdownOutput').value = '';
@@ -97,7 +97,7 @@ function generateJSON() {
setParseStatus(`Parsed ${issues.length} issues from input.`); setParseStatus(`Parsed ${issues.length} issues from input.`);
} }
export function parseIssuesFromHtmlTable(htmlText) { export function parseIssuesFromHtmlTable(htmlText, options = {}) {
const parser = new DOMParser(); const parser = new DOMParser();
const doc = parser.parseFromString(htmlText, 'text/html'); const doc = parser.parseFromString(htmlText, 'text/html');
const table = doc.querySelector('table'); const table = doc.querySelector('table');
@@ -109,6 +109,10 @@ export function parseIssuesFromHtmlTable(htmlText) {
const rows = Array.from(table.querySelectorAll('tr')); const rows = Array.from(table.querySelectorAll('tr'));
const issues = []; const issues = [];
const parsedType = normalizeWhitespace(String(options.type || '')).toLowerCase();
const includeDescriptionCaveats = parsedType !== 'addressed';
const includeInlineCodeFormatting = parsedType !== 'addressed';
rows.forEach(row => { rows.forEach(row => {
const cells = Array.from(row.children).filter(cell => { const cells = Array.from(row.children).filter(cell => {
const tag = (cell.tagName || '').toUpperCase(); const tag = (cell.tagName || '').toUpperCase();
@@ -119,8 +123,11 @@ export function parseIssuesFromHtmlTable(htmlText) {
return; return;
} }
const leftCellText = extractCellText(cells[0]); const leftCellText = extractCellPlainText(cells[0]);
const issueDetails = extractIssueDetails(cells[1]); const issueDetails = extractIssueDetails(cells[1], {
includeCaveat: includeDescriptionCaveats,
includeInlineCode: includeInlineCodeFormatting
});
const rightCellText = issueDetails.description; const rightCellText = issueDetails.description;
const rightCellCaveat = issueDetails.caveat; const rightCellCaveat = issueDetails.caveat;
@@ -134,13 +141,9 @@ export function parseIssuesFromHtmlTable(htmlText) {
} }
const id = issueIdMatch[1].toUpperCase(); const id = issueIdMatch[1].toUpperCase();
const leftCellTrailingText = normalizeWhitespace( const leftCellTrailingText = normalizeWhitespace(leftCellText.replace(issueIdMatch[0], ''));
leftCellText.replace(issueIdMatch[0], '').replace(/^[-:;,.\s]+/, '') const metadataText = leftCellTrailingText.replace(/^[-:;,.\s]+/, '');
); const resolved = RESOLVED_LINE_PATTERN.test(metadataText) ? metadataText : '';
// Strip markdown formatting to check if this is resolved text
const plainTrailingText = stripMarkdownFormatting(leftCellTrailingText);
const resolved = RESOLVED_LINE_PATTERN.test(plainTrailingText) ? plainTrailingText : '';
const description = rightCellText; const description = rightCellText;
if (!description) { if (!description) {
@@ -151,39 +154,29 @@ export function parseIssuesFromHtmlTable(htmlText) {
id, id,
description, description,
resolved, resolved,
caveat: resolved ? '' : (rightCellCaveat || plainTrailingText) caveat: resolved ? '' : (rightCellCaveat || metadataText)
}); });
}); });
return issues; return issues;
} }
function stripMarkdownFormatting(text) { function extractIssueDetails(cell, options = {}) {
return String(text || '')
.replace(/\*\*(.*?)\*\*/g, '$1')
.replace(/\*(.*?)\*/g, '$1')
.replace(/`(.*?)`/g, '$1')
.replace(/~~(.*?)~~/g, '$1');
}
function extractIssueDetails(cell) {
const caveatNode = cell.querySelector('tt');
const caveat = caveatNode ? normalizeWhitespace(caveatNode.textContent || '') : '';
if (!caveatNode) {
return {
description: extractCellText(cell),
caveat: ''
};
}
const clone = cell.cloneNode(true); const clone = cell.cloneNode(true);
const cloneCaveatNode = clone.querySelector('tt'); const includeCaveat = options.includeCaveat !== false;
if (cloneCaveatNode) { const includeInlineCode = options.includeInlineCode !== false;
cloneCaveatNode.remove();
let caveat = '';
const firstTtNode = clone.querySelector('tt');
if (firstTtNode && includeCaveat) {
caveat = normalizeWhitespace(firstTtNode.textContent || '');
firstTtNode.remove();
} else if (firstTtNode) {
Array.from(clone.querySelectorAll('tt')).forEach(node => node.remove());
} }
let description = extractCellText(clone); let description = extractCellText(clone, { includeInlineCode });
description = description.replace(/^\(\s*\)\s*/, '').trim(); description = description.replace(/^\(\s*\)\s*/, '').trim();
return { return {
@@ -192,15 +185,19 @@ function extractIssueDetails(cell) {
}; };
} }
function extractCellText(cell) { function extractCellPlainText(cell) {
return normalizeWhitespace(cell.textContent || '');
}
function extractCellText(cell, options = {}) {
const clone = cell.cloneNode(true); const clone = cell.cloneNode(true);
return collectMarkdownBlocks(clone) return collectMarkdownBlocks(clone, 0, options)
.join('\n\n') .join('\n\n')
.replace(/\n{3,}/g, '\n\n') .replace(/\n{3,}/g, '\n\n')
.trim(); .trim();
} }
function collectMarkdownBlocks(root, nestedListDepth = 0) { function collectMarkdownBlocks(root, nestedListDepth = 0, options = {}) {
const blocks = []; const blocks = [];
let paragraphBuffer = ''; let paragraphBuffer = '';
@@ -235,7 +232,7 @@ function collectMarkdownBlocks(root, nestedListDepth = 0) {
if (tagName === 'UL' || tagName === 'OL') { if (tagName === 'UL' || tagName === 'OL') {
flushParagraph(); flushParagraph();
const listMarkdown = convertHtmlListToMarkdown(node, nestedListDepth); const listMarkdown = convertHtmlListToMarkdown(node, nestedListDepth, options);
if (listMarkdown) { if (listMarkdown) {
blocks.push(listMarkdown); blocks.push(listMarkdown);
} }
@@ -244,18 +241,18 @@ function collectMarkdownBlocks(root, nestedListDepth = 0) {
if (BLOCK_CONTAINER_TAGS.has(tagName)) { if (BLOCK_CONTAINER_TAGS.has(tagName)) {
flushParagraph(); flushParagraph();
blocks.push(...collectMarkdownBlocks(node, nestedListDepth)); blocks.push(...collectMarkdownBlocks(node, nestedListDepth, options));
return; return;
} }
paragraphBuffer = appendInlineText(paragraphBuffer, convertInlineNodeToMarkdown(node)); paragraphBuffer = appendInlineText(paragraphBuffer, convertInlineNodeToMarkdown(node, options));
}); });
flushParagraph(); flushParagraph();
return blocks; return blocks;
} }
function convertInlineNodeToMarkdown(node) { function convertInlineNodeToMarkdown(node, options = {}) {
if (!node || node.nodeType !== ELEMENT_NODE) { if (!node || node.nodeType !== ELEMENT_NODE) {
return ''; return '';
} }
@@ -270,25 +267,33 @@ function convertInlineNodeToMarkdown(node) {
} }
if (/\buicontrol\b/i.test(className)) { if (/\buicontrol\b/i.test(className)) {
const content = normalizeInlineMarkdown(renderInlineChildren(node)); const content = normalizeInlineMarkdown(renderInlineChildren(node, options));
return content ? `**${content}**` : ''; return content ? `**${content}**` : '';
} }
if (/\buserinput\b/i.test(className)) { if (/\buserinput\b/i.test(className)) {
const content = normalizeInlineMarkdown(renderInlineChildren(node)); const content = normalizeInlineMarkdown(renderInlineChildren(node, options));
return content ? `\`${content}\`` : ''; if (!content) {
return '';
}
return options.includeInlineCode === false ? content : `\`${content}\``;
} }
if (/\bsystemoutput\b/i.test(className)) { if (/\bsystemoutput\b/i.test(className)) {
const content = normalizeInlineMarkdown(renderInlineChildren(node)); const content = normalizeInlineMarkdown(renderInlineChildren(node, options));
return content ? `\`${content}\`` : ''; if (!content) {
return '';
}
return options.includeInlineCode === false ? content : `\`${content}\``;
} }
const tagName = (node.tagName || '').toUpperCase(); const tagName = (node.tagName || '').toUpperCase();
if (tagName === 'A' && /\bxref\b/i.test(className)) { if (tagName === 'A' && /\bxref\b/i.test(className)) {
const href = node.getAttribute('href') || ''; const href = node.getAttribute('href') || '';
const content = normalizeInlineMarkdown(renderInlineChildren(node)); const content = normalizeInlineMarkdown(renderInlineChildren(node, options));
return content && href ? `[${content}](${href})` : content; return content && href ? `[${content}](${href})` : content;
} }
@@ -297,19 +302,19 @@ function convertInlineNodeToMarkdown(node) {
} }
if (tagName === 'STRONG' || tagName === 'B') { if (tagName === 'STRONG' || tagName === 'B') {
const content = normalizeInlineMarkdown(renderInlineChildren(node)); const content = normalizeInlineMarkdown(renderInlineChildren(node, options));
return content ? `**${content}**` : ''; return content ? `**${content}**` : '';
} }
if (tagName === 'EM' || tagName === 'I') { if (tagName === 'EM' || tagName === 'I') {
const content = normalizeInlineMarkdown(renderInlineChildren(node)); const content = normalizeInlineMarkdown(renderInlineChildren(node, options));
return content ? `*${content}*` : ''; return content ? `*${content}*` : '';
} }
return renderInlineChildren(node); return renderInlineChildren(node, options);
} }
function renderInlineChildren(node) { function renderInlineChildren(node, options = {}) {
let output = ''; let output = '';
Array.from(node.childNodes).forEach(child => { Array.from(node.childNodes).forEach(child => {
@@ -322,7 +327,7 @@ function renderInlineChildren(node) {
return; return;
} }
output = appendInlineText(output, convertInlineNodeToMarkdown(child)); output = appendInlineText(output, convertInlineNodeToMarkdown(child, options));
}); });
return output; return output;
@@ -340,7 +345,7 @@ function extractMenuCascadeParts(node) {
.filter(Boolean); .filter(Boolean);
} }
function convertHtmlListToMarkdown(list, depth = 0) { function convertHtmlListToMarkdown(list, depth = 0, options = {}) {
const items = Array.from(list.children).filter(child => child.tagName && child.tagName.toUpperCase() === 'LI'); const items = Array.from(list.children).filter(child => child.tagName && child.tagName.toUpperCase() === 'LI');
if (items.length === 0) { if (items.length === 0) {
return ''; return '';
@@ -348,13 +353,13 @@ function convertHtmlListToMarkdown(list, depth = 0) {
const isOrdered = list.tagName.toUpperCase() === 'OL'; const isOrdered = list.tagName.toUpperCase() === 'OL';
return items return items
.map((item, index) => convertListItemToMarkdown(item, isOrdered ? `${index + 1}.` : '-', depth)) .map((item, index) => convertListItemToMarkdown(item, isOrdered ? `${index + 1}.` : '-', depth, options))
.filter(Boolean) .filter(Boolean)
.join('\n'); .join('\n');
} }
function convertListItemToMarkdown(item, marker, depth) { function convertListItemToMarkdown(item, marker, depth, options = {}) {
const blocks = collectMarkdownBlocks(item, depth + 1); const blocks = collectMarkdownBlocks(item, depth + 1, options);
const indent = ' '.repeat(depth); const indent = ' '.repeat(depth);
const continuationIndent = `${indent} `; const continuationIndent = `${indent} `;