[FEAT]: Make PDF Processing Faster with Template Caching

[Cubix33]

📝 Description

Right now, FireForm reads and calculates the layout of a PDF form from scratch every single time it runs. If a fire department fills out the exact same "Incident Report" 100 times a day, our system wastes time and processing power re-reading the same blank form 100 times.

💡 Rationale

We should save (or "cache") the blank form in the system's memory the very first time we see it. The next 99 times, FireForm will skip the slow hard-drive reading process and load the form instantly from memory. This will make the system much faster and allow it to handle more forms at once.

🛠️ Proposed solution

We can do this by adding a simple memory dictionary to our PDF filler code.

• Create the Cache: Add a dictionary (self._template_cache = {}) inside the Filler class to store the raw bytes of the PDFs.
• Update PDF Loading: In the fill_form function, check if the PDF is already in the cache.
  • If it's new, read the file from the hard drive and save it to the cache.
  • If it's already saved, load it instantly using PdfReader(fdata=...).
• Reset LLM Memory: Ensure self._json is reset to an empty dictionary {} at the start of main_loop in src/llm.py. This prevents older answers from crashing the system when doing multiple fast runs.

✅ Acceptance Criteria

• The "Double Run" Test: If we send the exact same form to the system twice in a row, the terminal prints [LOG] Template Cache Hit! on the second run.
• No Crashes: The system successfully processes the second run without throwing an AttributeError or memory state bug.
• Speed: The second run finishes noticeably faster than the first run because it skips reading the file from the disk.