Optimize get-clip SRT lookup with LRU cache and time-based indexing

[Copilot]

The get-clip handler was loading SRT files from S3 on every request and filtering all entries for each segment, resulting in O(segments × entries) complexity. For clips with multiple segments and large SRT files (hundreds of entries), this created unnecessary I/O and computational overhead.

Changes:

• LRU cache with 50-entry limit: Caches parsed and indexed SRT data to prevent repeated S3 downloads and re-parsing. Evicts oldest entries when full to prevent unbounded memory growth in long-running Lambda containers.

• Time-based bucket index: Groups SRT entries into 10-second buckets. Lookup now queries only relevant buckets instead of filtering all entries, reducing complexity to O(segments × entries_per_bucket).

• Reuses existing infrastructure: Leverages the loadTranscript utility which already provides S3-level caching.

// Before: Filter all entries for each segmentconstrelevantEntries=srtEntries.filter(entry=>{constentryStart=timeToSeconds(entry.startTime);constentryEnd=timeToSeconds(entry.endTime);returnentryStart<segEnd&&entryEnd>segStart;});// After: Query indexed buckets for candidates, then filterconstrelevantEntries=findRelevantEntries(srtData.index,segStart,segEnd);

Module constants MAX_CACHE_SIZE and BUCKET_SIZE allow easy tuning based on operational metrics.

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.

[Copilot]

The get-clip handler was loading SRT files from S3 on every request and filtering all entries for each segment, resulting in O(segments × entries) complexity. For clips with multiple segments and large SRT files (hundreds of entries), this created unnecessary I/O and computational overhead.

Changes:

• LRU cache with 50-entry limit: Caches parsed and indexed SRT data to prevent repeated S3 downloads and re-parsing. Evicts oldest entries when full to prevent unbounded memory growth in long-running Lambda containers.

• Time-based bucket index: Groups SRT entries into 10-second buckets. Lookup now queries only relevant buckets instead of filtering all entries, reducing complexity to O(segments × entries_per_bucket).

• Reuses existing infrastructure: Leverages the loadTranscript utility which already provides S3-level caching.

// Before: Filter all entries for each segmentconstrelevantEntries=srtEntries.filter(entry=>{constentryStart=timeToSeconds(entry.startTime);constentryEnd=timeToSeconds(entry.endTime);returnentryStart<segEnd&&entryEnd>segStart;});// After: Query indexed buckets for candidates, then filterconstrelevantEntries=findRelevantEntries(srtData.index,segStart,segEnd);

Module constants MAX_CACHE_SIZE and BUCKET_SIZE allow easy tuning based on operational metrics.

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.

[Copilot]

The get-clip handler was loading SRT files from S3 on every request and filtering all entries for each segment, resulting in O(segments × entries) complexity. For clips with multiple segments and large SRT files (hundreds of entries), this created unnecessary I/O and computational overhead.

Changes:

• LRU cache with 50-entry limit: Caches parsed and indexed SRT data to prevent repeated S3 downloads and re-parsing. Evicts oldest entries when full to prevent unbounded memory growth in long-running Lambda containers.

• Time-based bucket index: Groups SRT entries into 10-second buckets. Lookup now queries only relevant buckets instead of filtering all entries, reducing complexity to O(segments × entries_per_bucket).

• Reuses existing infrastructure: Leverages the loadTranscript utility which already provides S3-level caching.

// Before: Filter all entries for each segmentconstrelevantEntries=srtEntries.filter(entry=>{constentryStart=timeToSeconds(entry.startTime);constentryEnd=timeToSeconds(entry.endTime);returnentryStart<segEnd&&entryEnd>segStart;});// After: Query indexed buckets for candidates, then filterconstrelevantEntries=findRelevantEntries(srtData.index,segStart,segEnd);

Module constants MAX_CACHE_SIZE and BUCKET_SIZE allow easy tuning based on operational metrics.

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.

[Copilot]

The get-clip handler was loading SRT files from S3 on every request and filtering all entries for each segment, resulting in O(segments × entries) complexity. For clips with multiple segments and large SRT files (hundreds of entries), this created unnecessary I/O and computational overhead.

Changes:

• LRU cache with 50-entry limit: Caches parsed and indexed SRT data to prevent repeated S3 downloads and re-parsing. Evicts oldest entries when full to prevent unbounded memory growth in long-running Lambda containers.

• Time-based bucket index: Groups SRT entries into 10-second buckets. Lookup now queries only relevant buckets instead of filtering all entries, reducing complexity to O(segments × entries_per_bucket).

• Reuses existing infrastructure: Leverages the loadTranscript utility which already provides S3-level caching.

// Before: Filter all entries for each segmentconstrelevantEntries=srtEntries.filter(entry=>{constentryStart=timeToSeconds(entry.startTime);constentryEnd=timeToSeconds(entry.endTime);returnentryStart<segEnd&&entryEnd>segStart;});// After: Query indexed buckets for candidates, then filterconstrelevantEntries=findRelevantEntries(srtData.index,segStart,segEnd);

Module constants MAX_CACHE_SIZE and BUCKET_SIZE allow easy tuning based on operational metrics.

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.

[Copilot]

The get-clip handler was loading SRT files from S3 on every request and filtering all entries for each segment, resulting in O(segments × entries) complexity. For clips with multiple segments and large SRT files (hundreds of entries), this created unnecessary I/O and computational overhead.

Changes:

• LRU cache with 50-entry limit: Caches parsed and indexed SRT data to prevent repeated S3 downloads and re-parsing. Evicts oldest entries when full to prevent unbounded memory growth in long-running Lambda containers.

• Time-based bucket index: Groups SRT entries into 10-second buckets. Lookup now queries only relevant buckets instead of filtering all entries, reducing complexity to O(segments × entries_per_bucket).

• Reuses existing infrastructure: Leverages the loadTranscript utility which already provides S3-level caching.

// Before: Filter all entries for each segmentconstrelevantEntries=srtEntries.filter(entry=>{constentryStart=timeToSeconds(entry.startTime);constentryEnd=timeToSeconds(entry.endTime);returnentryStart<segEnd&&entryEnd>segStart;});// After: Query indexed buckets for candidates, then filterconstrelevantEntries=findRelevantEntries(srtData.index,segStart,segEnd);

Module constants MAX_CACHE_SIZE and BUCKET_SIZE allow easy tuning based on operational metrics.

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.

[Copilot]

The get-clip handler was loading SRT files from S3 on every request and filtering all entries for each segment, resulting in O(segments × entries) complexity. For clips with multiple segments and large SRT files (hundreds of entries), this created unnecessary I/O and computational overhead.

Changes:

• LRU cache with 50-entry limit: Caches parsed and indexed SRT data to prevent repeated S3 downloads and re-parsing. Evicts oldest entries when full to prevent unbounded memory growth in long-running Lambda containers.

• Time-based bucket index: Groups SRT entries into 10-second buckets. Lookup now queries only relevant buckets instead of filtering all entries, reducing complexity to O(segments × entries_per_bucket).

• Reuses existing infrastructure: Leverages the loadTranscript utility which already provides S3-level caching.

// Before: Filter all entries for each segmentconstrelevantEntries=srtEntries.filter(entry=>{constentryStart=timeToSeconds(entry.startTime);constentryEnd=timeToSeconds(entry.endTime);returnentryStart<segEnd&&entryEnd>segStart;});// After: Query indexed buckets for candidates, then filterconstrelevantEntries=findRelevantEntries(srtData.index,segStart,segEnd);

Module constants MAX_CACHE_SIZE and BUCKET_SIZE allow easy tuning based on operational metrics.

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.

[Copilot]

The get-clip handler was loading SRT files from S3 on every request and filtering all entries for each segment, resulting in O(segments × entries) complexity. For clips with multiple segments and large SRT files (hundreds of entries), this created unnecessary I/O and computational overhead.

Changes:

• LRU cache with 50-entry limit: Caches parsed and indexed SRT data to prevent repeated S3 downloads and re-parsing. Evicts oldest entries when full to prevent unbounded memory growth in long-running Lambda containers.

• Time-based bucket index: Groups SRT entries into 10-second buckets. Lookup now queries only relevant buckets instead of filtering all entries, reducing complexity to O(segments × entries_per_bucket).

• Reuses existing infrastructure: Leverages the loadTranscript utility which already provides S3-level caching.

// Before: Filter all entries for each segmentconstrelevantEntries=srtEntries.filter(entry=>{constentryStart=timeToSeconds(entry.startTime);constentryEnd=timeToSeconds(entry.endTime);returnentryStart<segEnd&&entryEnd>segStart;});// After: Query indexed buckets for candidates, then filterconstrelevantEntries=findRelevantEntries(srtData.index,segStart,segEnd);

Module constants MAX_CACHE_SIZE and BUCKET_SIZE allow easy tuning based on operational metrics.

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.

[Copilot]

The get-clip handler was loading SRT files from S3 on every request and filtering all entries for each segment, resulting in O(segments × entries) complexity. For clips with multiple segments and large SRT files (hundreds of entries), this created unnecessary I/O and computational overhead.

Changes:

• LRU cache with 50-entry limit: Caches parsed and indexed SRT data to prevent repeated S3 downloads and re-parsing. Evicts oldest entries when full to prevent unbounded memory growth in long-running Lambda containers.

• Time-based bucket index: Groups SRT entries into 10-second buckets. Lookup now queries only relevant buckets instead of filtering all entries, reducing complexity to O(segments × entries_per_bucket).

• Reuses existing infrastructure: Leverages the loadTranscript utility which already provides S3-level caching.

// Before: Filter all entries for each segmentconstrelevantEntries=srtEntries.filter(entry=>{constentryStart=timeToSeconds(entry.startTime);constentryEnd=timeToSeconds(entry.endTime);returnentryStart<segEnd&&entryEnd>segStart;});// After: Query indexed buckets for candidates, then filterconstrelevantEntries=findRelevantEntries(srtData.index,segStart,segEnd);

Module constants MAX_CACHE_SIZE and BUCKET_SIZE allow easy tuning based on operational metrics.

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.