You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: wiki/ARCHITECTURE.md
+4-4Lines changed: 4 additions & 4 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -1,6 +1,6 @@
1
1
# FireForm Architecture
2
2
3
-
FireForm is a system designed to automate the process of filling out PDF forms using information extracted from transcriptions (e.g., voice recordings) via an LLM.
3
+
FireForm is a system designed to automate the process of filling out PDF forms using information extracted from transcriptions via an LLM.
4
4
5
5
## Pipeline Description
6
6
@@ -11,7 +11,7 @@ The FireForm pipeline follows a structured flow from receiving raw user input to
11
11
1.**Template Preparation**: A standard PDF is processed to ensure it has interactive form fields.
12
12
2.**Input Acquisition**: The system receives a transcript (text) and a list of target fields to extract.
13
13
3.**LLM Extraction**: An LLM (via Ollama) processes the transcript to find values for each target field.
14
-
4.**Form Filling**: extracted values are mapped and written into the PDF template.
14
+
4.**Form Filling**: Extracted values are mapped and written into the PDF template.
15
15
5.**Output Generation**: A new, filled PDF is saved with a timestamped filename.
16
16
17
17
### Component Interaction
@@ -55,7 +55,7 @@ sequenceDiagram
55
55
### Key Components
56
56
57
57
-**`FileManipulator`**: The high-level orchestrator that manages the overall process.
58
-
-**`LLM`**: Handles prompt engineering and communication with the local Ollama instance (typically using the Mistral model). It extracts structured data from unstructured text.
59
-
-**`Filler`**: Handles the low-level PDF manipulation, ensuring that extracted data is placed into the correct form fields based on their visual order.
58
+
-**`LLM`**: Handles prompt engineering and communication with the local Ollama instance. It extracts structured data from unstructured text.
59
+
-**`Filler`**: Handles the low-level PDF manipulation, ensuring that extracted data is placed into the correct form fields.
60
60
-**`commonforms`**: A dependency used to prepare standard PDFs for filling.
61
61
-**`pdfrw` / `pypdf`**: Libraries used for reading and writing PDF metadata and form field values.
0 commit comments