fix: better macro analysis

This commit is contained in:
2026-02-13 16:27:35 +01:00
parent 1aa44e675d
commit 00e8255cd7
2 changed files with 43 additions and 2 deletions

View File

@@ -146,7 +146,8 @@ export class ImportAnalysisEngine {
private static readonly SHORTCODE_REGEX = /(?<!\[)\[(\w+)([^\]]*?)(?:\s*\/)?\](?!\])/g;
// Regex to extract individual parameters from shortcode
private static readonly PARAM_REGEX = /(\w+)=["']([^"']*?)["']/g;
// Supports: key="value", key='value', and key=value (unquoted)
private static readonly PARAM_REGEX = /(\w+)=(?:"([^"]*)"|'([^']*)'|([^\s\]"']+))/g;
constructor() {
this.turndown = new TurndownService({
@@ -574,6 +575,7 @@ export class ImportAnalysisEngine {
/**
* Parse parameters from a shortcode parameter string.
* Supports: key="value", key='value', and key=value (unquoted)
*/
private parseShortcodeParams(paramString: string): Record<string, string> {
const params: Record<string, string> = {};
@@ -583,7 +585,10 @@ export class ImportAnalysisEngine {
let match;
while ((match = ImportAnalysisEngine.PARAM_REGEX.exec(paramString)) !== null) {
params[match[1]] = match[2];
const key = match[1];
// Value is in group 2 (double-quoted), 3 (single-quoted), or 4 (unquoted)
const value = match[2] ?? match[3] ?? match[4] ?? '';
params[key] = value;
}
return params;

View File

@@ -709,6 +709,42 @@ describe('ImportAnalysisEngine', () => {
expect(youtubeMacro?.usages[0].params.id).toBe('singlequoted');
});
it('should handle shortcodes with unquoted parameters', async () => {
setupDbReturns([], [], []);
const wxrData = createWxrData({
posts: [createWxrPost({
content: '[youtube id=abc123def4g autoplay=true][gallery columns=4]',
})],
});
const report = await engine.analyzeWxr(wxrData, '/test.xml');
const youtubeMacro = report.macros.discovered.find(m => m.name === 'youtube');
expect(youtubeMacro?.usages[0].params.id).toBe('abc123def4g');
expect(youtubeMacro?.usages[0].params.autoplay).toBe('true');
const galleryMacro = report.macros.discovered.find(m => m.name === 'gallery');
expect(galleryMacro?.usages[0].params.columns).toBe('4');
});
it('should handle mixed quoted and unquoted parameters', async () => {
setupDbReturns([], [], []);
const wxrData = createWxrData({
posts: [createWxrPost({
content: '[video src="http://example.com/video.mp4" autoplay=true width=640]',
})],
});
const report = await engine.analyzeWxr(wxrData, '/test.xml');
const videoMacro = report.macros.discovered.find(m => m.name === 'video');
expect(videoMacro?.usages[0].params.src).toBe('http://example.com/video.mp4');
expect(videoMacro?.usages[0].params.autoplay).toBe('true');
expect(videoMacro?.usages[0].params.width).toBe('640');
});
it('should not detect our internal macro format as WordPress shortcodes', async () => {
setupDbReturns([], [], []);