Merge pull request #196 from cloud-atlas-ai/feature/configurable-video-call-urls

Feature: Field Extraction

Author: muness

README.md

@@ -13,8 +13,9 @@ export interface IEvent {
 	description: string; // Detailed description of the event
 	format: Calendar["format"]; // Format preference for the event
 	location: string; // Physical location where the event takes place, if applicable
-	callUrl: string; // URL for joining online meetings/calls associated with the event
-	callType: string; // Type of online meeting (e.g., Zoom, Skype, etc.)
+	callUrl: string; // URL for joining online meetings/calls associated with the event (backward compatibility)
+	callType: string; // Type of online meeting (e.g., Zoom, Skype, etc.) (backward compatibility)
+	extractedFields: Record<string, string[]>; // Generic field extraction results
 	organizer: IOrganizer; // Email of the organizer of the event
   attendees: IAttendee[]; // Array of attendees
   eventType: string; // Type of event (e.g., one-off, recurring, recurring override)

src/IEvent.ts

@@ -3,30 +3,94 @@ import { tz } from 'moment-timezone';
 import { moment } from "obsidian";
 import { WINDOWS_TO_IANA_TIMEZONES } from './generated/windowsTimezones';
 
-export function extractMeetingInfo(e: any): { callUrl: string, callType: string } {
+import { FieldExtractionPattern } from './settings/ICSSettings';
 
-  // Check for Google Meet conference data
-  if (e["GOOGLE-CONFERENCE"]) {
-    return { callUrl: e["GOOGLE-CONFERENCE"], callType: 'Google Meet' };
+export function extractFields(e: any, patterns?: FieldExtractionPattern[]): Record<string, string[]> {
+  // If patterns not provided or empty, return empty object
+  if (!patterns || patterns.length === 0) {
+    return {};
   }
-  // Check if the location contains a Zoom link
-  if (e.location && e.location.includes('zoom.us')) {
-    return { callUrl: e.location, callType: 'Zoom' };
+
+  const extractedFields: Record<string, string[]> = {};
+
+  // Sort patterns by priority (lower numbers = higher priority)
+  const sortedPatterns = patterns.sort((a, b) => a.priority - b.priority);
+
+  for (const pattern of sortedPatterns) {
+    const matches = findPatternMatches(e, pattern);
+    if (matches.length > 0) {
+      const fieldName = pattern.extractedFieldName;
+      if (!extractedFields[fieldName]) {
+        extractedFields[fieldName] = [];
+      }
+      extractedFields[fieldName].push(...matches);
+    }
+  }
+
+  // Deduplicate all extracted fields
+  for (const fieldName in extractedFields) {
+    extractedFields[fieldName] = [...new Set(extractedFields[fieldName])];
+  }
+
+  return extractedFields;
+}
+
+function findPatternMatches(e: any, pattern: FieldExtractionPattern): string[] {
+  const matches: string[] = [];
+
+  // Special handling for Google Meet conference data
+  if (pattern.pattern === "GOOGLE-CONFERENCE" && e["GOOGLE-CONFERENCE"]) {
+    matches.push(e["GOOGLE-CONFERENCE"]);
+    return matches;
+  }
+
+  // Check location field
+  if (e.location) {
+    const locationMatches = matchTextForPattern(e.location, pattern);
+    matches.push(...locationMatches);
   }
+
+  // Check description field
   if (e.description) {
-    const skypeMatch = e.description.match(/https:\/\/join.skype.com\/[a-zA-Z0-9]+/);
-    if (skypeMatch) {
-      return { callUrl: skypeMatch[0], callType: 'Skype' };
-    }
+    const descriptionMatches = matchTextForPattern(e.description, pattern);
+    matches.push(...descriptionMatches);
+  }
+
+  return matches;
+}
+
+function matchTextForPattern(text: string, pattern: FieldExtractionPattern): string[] {
+  const matches: string[] = [];
 
-    const teamsMatch = e.description.match(/(https:\/\/teams\.microsoft\.com\/l\/meetup-join\/[^>]+)/);
-    if (teamsMatch) {
-      return { callUrl: teamsMatch[0], callType: 'Microsoft Teams' };
+  try {
+    if (pattern.matchType === 'contains') {
+      if (text.includes(pattern.pattern)) {
+        // For contains match, try to extract URLs from the text
+        const urlMatches = text.match(/https?:\/\/[^\s<>"]+/g);
+        if (urlMatches) {
+          matches.push(...urlMatches);
+        } else {
+          // If no URLs found, return the original text
+          matches.push(text);
+        }
+      }
+    } else if (pattern.matchType === 'regex') {
+      const regex = new RegExp(pattern.pattern, 'g'); // Use global flag to find all matches
+      let match;
+      while ((match = regex.exec(text)) !== null) {
+        // If regex has capture groups, use the first group, otherwise use full match
+        matches.push(match[1] || match[0]);
+      }
     }
+  } catch {
+    // Skip invalid regex patterns
+    console.warn(`Invalid regex pattern: ${pattern.pattern}`);
   }
-  return { callUrl: null, callType: null };
+
+  return matches;
 }
 
+
 function applyRecurrenceDateAndTimezone(originalDate: Date, currentDate: Date, tzid: string): Date {
   const originalMoment = tz(originalDate, tzid);
 

src/icalUtils.ts

@@ -7,6 +7,7 @@ import {
   Calendar,
   ICSSettings,
   DEFAULT_SETTINGS,
+  DEFAULT_FIELD_EXTRACTION_PATTERNS,
 } from "./settings/ICSSettings";
 
 import ICSSettingsTab from "./settings/ICSSettingsTab";
@@ -19,7 +20,7 @@ import {
   Plugin,
   request
 } from 'obsidian';
-import { parseIcs, filterMatchingEvents, extractMeetingInfo } from './icalUtils';
+import { parseIcs, filterMatchingEvents, extractFields } from './icalUtils';
 import { IEvent } from './IEvent';
 import { DateNormalizer, FlexibleDateInput } from './DateNormalizer';
 
@@ -135,7 +136,17 @@ export default class ICSPlugin extends Plugin {
 
       try {
         dateEvents.forEach((e) => {
-          const { callUrl, callType } = extractMeetingInfo(e);
+          const patterns = this.data.fieldExtraction?.enabled ? this.data.fieldExtraction.patterns : [];
+          const extractedFields = extractFields(e, patterns);
+
+          // Backward compatibility: extract first Video Call URL and type
+          // Support both old singular and new plural field names
+          const videoCallUrls = extractedFields['Video Call URLs'] || extractedFields['Video Call URL'] || [];
+          const callUrl = videoCallUrls.length > 0 ? videoCallUrls[0] : null;
+
+          // For callType, we could derive it from the pattern name, but since we're going generic,
+          // let's just use "Video Call" as a generic type when we have a URL
+          const callType = callUrl ? "Video Call" : null;
 
           const event: IEvent = {
             utime: moment(e.start).format('X'),
@@ -152,6 +163,7 @@ export default class ICSPlugin extends Plugin {
             location: e.location ? e.location : null,
             callUrl: callUrl,
             callType: callType,
+            extractedFields: extractedFields,
             eventType: e.eventType,
             organizer: { email: e.organizer?.val?.substring(7) || null, name: e.organizer?.params?.CN || null },
             attendees: e.attendee ? (Array.isArray(e.attendee) ? e.attendee : [e.attendee]).map(attendee => ({
@@ -259,6 +271,45 @@ export default class ICSPlugin extends Plugin {
 
   async loadSettings() {
     this.data = Object.assign({}, DEFAULT_SETTINGS, await this.loadData());
+
+    // Migration: migrate from old videoCallExtraction to new fieldExtraction
+    let needsSave = false;
+
+    // If old videoCallExtraction exists, migrate it to fieldExtraction
+    if ((this.data as any).videoCallExtraction) {
+      const oldSettings = (this.data as any).videoCallExtraction;
+      this.data.fieldExtraction = {
+        enabled: oldSettings.enabled !== false,
+        patterns: DEFAULT_FIELD_EXTRACTION_PATTERNS
+      };
+      delete (this.data as any).videoCallExtraction;
+      needsSave = true;
+    }
+
+    // Ensure fieldExtraction settings exist and are hydrated
+    if (!this.data.fieldExtraction) {
+      this.data.fieldExtraction = {
+        enabled: true,
+        patterns: [...DEFAULT_FIELD_EXTRACTION_PATTERNS]
+      };
+      needsSave = true;
+    } else {
+      // Ensure patterns array exists
+      if (!this.data.fieldExtraction.patterns) {
+        this.data.fieldExtraction.patterns = [...DEFAULT_FIELD_EXTRACTION_PATTERNS];
+        needsSave = true;
+      }
+
+      // Ensure enabled field exists
+      if (this.data.fieldExtraction.enabled === undefined) {
+        this.data.fieldExtraction.enabled = true;
+        needsSave = true;
+      }
+    }
+
+    if (needsSave) {
+      await this.saveData(this.data);
+    }
   }
 
   async saveSettings() {

src/main.ts

@@ -1,9 +1,21 @@
+export interface FieldExtractionPattern {
+    name: string;
+    pattern: string;
+    matchType: 'regex' | 'contains';
+    priority: number;
+    extractedFieldName: string;
+}
+
 export interface ICSSettings {
     format: {
         timeFormat: string
         dataViewSyntax: boolean,
     },
     calendars: Record < string, Calendar > ;
+    fieldExtraction: {
+        enabled: boolean;
+        patterns: FieldExtractionPattern[];
+    };
 }
 
 export interface Calendar {
@@ -43,11 +55,46 @@ export const DEFAULT_CALENDAR_FORMAT = {
     showTransparentEvents: false
 };
 
+export const DEFAULT_FIELD_EXTRACTION_PATTERNS: FieldExtractionPattern[] = [
+    {
+        name: "Google Meet",
+        pattern: "GOOGLE-CONFERENCE",
+        matchType: "contains",
+        priority: 1,
+        extractedFieldName: "Video Call URLs"
+    },
+    {
+        name: "Zoom",
+        pattern: "zoom.us",
+        matchType: "contains",
+        priority: 2,
+        extractedFieldName: "Video Call URLs"
+    },
+    {
+        name: "Skype",
+        pattern: "https:\\/\\/join\\.skype\\.com\\/[a-zA-Z0-9]+",
+        matchType: "regex",
+        priority: 3,
+        extractedFieldName: "Video Call URLs"
+    },
+    {
+        name: "Microsoft Teams",
+        pattern: "https:\\/\\/teams\\.microsoft\\.com\\/l\\/meetup-join\\/[^>]+",
+        matchType: "regex",
+        priority: 4,
+        extractedFieldName: "Video Call URLs"
+    }
+];
+
 export const DEFAULT_SETTINGS: ICSSettings = {
     format: {
         timeFormat: "HH:mm",
         dataViewSyntax: false,
     },
     calendars: {
+    },
+    fieldExtraction: {
+        enabled: true,
+        patterns: DEFAULT_FIELD_EXTRACTION_PATTERNS
     }
 };