spsa update

Author: dwillis

python-parsers/pdf_extractor.py

@@ -31,53 +31,24 @@ def extract_results_with_llm(pdf_path, county, model_name):
 4. Include rows with 0 votes - do not skip them
 5. Write-ins should have empty party field and candidate must be "Write-ins" (plural)
 6. Registered Voters and Ballots Cast have empty party and candidate fields
+7. Double-quote values with commas in them
 
 Example format:
 county,precinct,office,district,party,candidate,votes,absentee,early_voting,election_day
-Panola,1,Registered Voters,,,,1972,,,
-Panola,2,Registered Voters,,,,2332,,,
-Panola,3,Registered Voters,,,,1453,,,
-Panola,5,Registered Voters,,,,505,,,
-Panola,7,Registered Voters,,,,898,,,
-Panola,8,Registered Voters,,,,282,,,
-Panola,9,Registered Voters,,,,788,,,
-Panola,10,Registered Voters,,,,368,,,
-Panola,12,Registered Voters,,,,439,,,
-Panola,13,Registered Voters,,,,363,,,
-Panola,14,Registered Voters,,,,508,,,
-Panola,18,Registered Voters,,,,2240,,,
-Panola,19,Registered Voters,,,,380,,,
-Panola,20,Registered Voters,,,,360,,,
-Panola,22,Registered Voters,,,,465,,,
-Panola,26,Registered Voters,,,,177,,,
-Panola,27,Registered Voters,,,,1504,,,
-Panola,28,Registered Voters,,,,2518,,,
-Panola,29,Registered Voters,,,,365,,,
-Panola,1,Ballots Cast,,,,1264,56,954,254
-Panola,2,Ballots Cast,,,,1489,50,1132,307
-Panola,3,Ballots Cast,,,,986,19,639,328
-Panola,5,Ballots Cast,,,,390,6,227,157
-Panola,7,Ballots Cast,,,,590,12,280,298
-Panola,8,Ballots Cast,,,,181,1,93,87
-Panola,9,Ballots Cast,,,,497,17,190,290
-Panola,10,Ballots Cast,,,,229,7,134,88
-Panola,12,Ballots Cast,,,,339,16,178,145
-Panola,13,Ballots Cast,,,,230,8,124,98
-Panola,14,Ballots Cast,,,,348,5,215,128
-Panola,18,Ballots Cast,,,,1556,52,1063,441
-Panola,19,Ballots Cast,,,,263,6,128,129
-Panola,20,Ballots Cast,,,,266,2,174,90
-Panola,22,Ballots Cast,,,,256,8,102,146
-Panola,26,Ballots Cast,,,,127,2,60,65
-Panola,27,Ballots Cast,,,,777,27,535,215
-Panola,28,Ballots Cast,,,,1474,49,1117,308
-Panola,29,Ballots Cast,,,,238,3,104,131
-Panola,1,President,,REP,Donald J. Trump,1021,39,787,195
-Panola,2,President,,REP,Donald J. Trump,1182,35,937,210
-Panola,3,President,,REP,Donald J. Trump,816,16,545,255
-Panola,5,President,,REP,Donald J. Trump,368,4,220,144
-Panola,7,President,,REP,Donald J. Trump,495,8,249,238
-Panola,8,President,,REP,Donald J. Trump,154,0,80,74
+Scurry,Precinct 5,Registered Voters,,,,98,,,
+Scurry,Precinct 5,Ballots Cast,,,,73,0,65,8
+Scurry,Precinct 5,President,,REP,Donald J. Trump,60,0,54,6
+Scurry,Precinct 5,President,,DEM,Kamala D. Harris,13,0,11,2
+Scurry,Precinct 5,President,,LIB,Chase Oliver,0,0,0,0
+Scurry,Precinct 5,President,,GRN,Jill Stein,0,0,0,0
+Scurry,Precinct 5,President,,,Write-ins,0,0,0,0
+Scurry,Precinct 5,U.S. Senate,,REP,Ted Cruz,57,0,51,6
+Scurry,Precinct 5,U.S. Senate,,DEM,Colin Allred,14,0,12,2
+Scurry,Precinct 5,U.S. Senate,,LIB,Ted Brown,0,0,0,0
+Scurry,Precinct 5,U.S. Senate,,,Write-ins,0,0,0,0
+Scurry,Precinct 5,U.S. House,19,REP,Jodey C. Arrington,62,0,55,7
+Scurry,Precinct 5,U.S. House,19,LIB,Bernard Johnson,3,0,2,1
+Scurry,Precinct 5,U.S. House,19,IND,Nathan Lewis,3,0,3,0
 
 Extract ALL precinct-level results from the PDF following this exact format. Process every page of the PDF.
 Return ONLY the CSV data, no explanations or header row.