chore: add .gitignore

2026-03-05 00:37:34 +01:00
parent da34bdc8d7
commit d2e0fec97d
2087 changed files with 1485338 additions and 14 deletions
--- a/output/reports/27_23134370/CRO_23134370_report.json
+++ b/output/reports/27_23134370/CRO_23134370_report.json
@@ -0,0 +1,39 @@
+{
+  "source_file": "CRO 23134370.pdf",
+  "total_replacements": 16,
+  "regex_replacements": 7,
+  "ner_replacements": 5,
+  "sweep_replacements": 4,
+  "entities_found": [
+    {
+      "original": "Jackson Pratt",
+      "replacement": "[PERSONNE_1]",
+      "source": "ner",
+      "score": 0.9939177632331848
+    },
+    {
+      "original": "MARIE NAGOUAS",
+      "replacement": "[PERSONNE_2]",
+      "source": "ner",
+      "score": 0.955693781375885
+    },
+    {
+      "original": "Docteur M. [PATIENT_1",
+      "replacement": "[PERSONNE_3]",
+      "source": "ner",
+      "score": 0.9558792114257812
+    },
+    {
+      "original": "Salvador Allende",
+      "replacement": "[PERSONNE_4]",
+      "source": "ner",
+      "score": 0.9931547045707703
+    },
+    {
+      "original": "LEVERGE Jessica",
+      "replacement": "[PERSONNE_5]",
+      "source": "ner",
+      "score": 0.9779495000839233
+    }
+  ]
+}
--- a/output/reports/27_23134370/trackare-BA060895-23134370_BA060895_23134370_report.json
+++ b/output/reports/27_23134370/trackare-BA060895-23134370_BA060895_23134370_report.json
@@ -0,0 +1,633 @@
+{
+  "source_file": "trackare-BA060895-23134370_BA060895_23134370.pdf",
+  "total_replacements": 1048,
+  "regex_replacements": 183,
+  "ner_replacements": 71,
+  "sweep_replacements": 794,
+  "entities_found": [
+    {
+      "original": "23134370",
+      "replacement": "[EPISODE_1]",
+      "source": "regex",
+      "category": "episode"
+    },
+    {
+      "original": "23134370",
+      "replacement": "[EPISODE_1]",
+      "source": "regex",
+      "category": "episode"
+    },
+    {
+      "original": "23134370",
+      "replacement": "[EPISODE_1]",
+      "source": "regex",
+      "category": "episode"
+    },
+    {
+      "original": "23134370",
+      "replacement": "[EPISODE_1]",
+      "source": "regex",
+      "category": "episode"
+    },
+    {
+      "original": "23134370",
+      "replacement": "[EPISODE_1]",
+      "source": "regex",
+      "category": "episode"
+    },
+    {
+      "original": "23134370",
+      "replacement": "[EPISODE_1]",
+      "source": "regex",
+      "category": "episode"
+    },
+    {
+      "original": "23134370",
+      "replacement": "[EPISODE_1]",
+      "source": "regex",
+      "category": "episode"
+    },
+    {
+      "original": "23134370",
+      "replacement": "[EPISODE_1]",
+      "source": "regex",
+      "category": "episode"
+    },
+    {
+      "original": "23134370",
+      "replacement": "[EPISODE_1]",
+      "source": "regex",
+      "category": "episode"
+    },
+    {
+      "original": "23134370",
+      "replacement": "[EPISODE_1]",
+      "source": "regex",
+      "category": "episode"
+    },
+    {
+      "original": "23134370",
+      "replacement": "[EPISODE_1]",
+      "source": "regex",
+      "category": "episode"
+    },
+    {
+      "original": "23134370",
+      "replacement": "[EPISODE_1]",
+      "source": "regex",
+      "category": "episode"
+    },
+    {
+      "original": "23134370",
+      "replacement": "[EPISODE_1]",
+      "source": "regex",
+      "category": "episode"
+    },
+    {
+      "original": "23134370",
+      "replacement": "[EPISODE_1]",
+      "source": "regex",
+      "category": "episode"
+    },
+    {
+      "original": "23134370",
+      "replacement": "[EPISODE_1]",
+      "source": "regex",
+      "category": "episode"
+    },
+    {
+      "original": "23134370",
+      "replacement": "[EPISODE_1]",
+      "source": "regex",
+      "category": "episode"
+    },
+    {
+      "original": "23134370",
+      "replacement": "[EPISODE_1]",
+      "source": "regex",
+      "category": "episode"
+    },
+    {
+      "original": "23134370",
+      "replacement": "[EPISODE_1]",
+      "source": "regex",
+      "category": "episode"
+    },
+    {
+      "original": "23134370",
+      "replacement": "[EPISODE_1]",
+      "source": "regex",
+      "category": "episode"
+    },
+    {
+      "original": "23134370",
+      "replacement": "[EPISODE_1]",
+      "source": "regex",
+      "category": "episode"
+    },
+    {
+      "original": "23134370",
+      "replacement": "[EPISODE_1]",
+      "source": "regex",
+      "category": "episode"
+    },
+    {
+      "original": "23134370",
+      "replacement": "[EPISODE_1]",
+      "source": "regex",
+      "category": "episode"
+    },
+    {
+      "original": "23134370",
+      "replacement": "[EPISODE_1]",
+      "source": "regex",
+      "category": "episode"
+    },
+    {
+      "original": "23134370",
+      "replacement": "[EPISODE_1]",
+      "source": "regex",
+      "category": "episode"
+    },
+    {
+      "original": "23134370",
+      "replacement": "[EPISODE_1]",
+      "source": "regex",
+      "category": "episode"
+    },
+    {
+      "original": "23134370",
+      "replacement": "[EPISODE_1]",
+      "source": "regex",
+      "category": "episode"
+    },
+    {
+      "original": "23134370",
+      "replacement": "[EPISODE_1]",
+      "source": "regex",
+      "category": "episode"
+    },
+    {
+      "original": "23134370",
+      "replacement": "[EPISODE_1]",
+      "source": "regex",
+      "category": "episode"
+    },
+    {
+      "original": "23134370",
+      "replacement": "[EPISODE_1]",
+      "source": "regex",
+      "category": "episode"
+    },
+    {
+      "original": "640000162",
+      "replacement": "[FINESS]",
+      "source": "regex",
+      "category": "finess"
+    },
+    {
+      "original": "99999",
+      "replacement": "[LIEU_NAISS_2]",
+      "source": "regex",
+      "category": "lieu_naissance"
+    },
+    {
+      "original": "99999",
+      "replacement": "[LIEU_NAISS_2]",
+      "source": "regex",
+      "category": "lieu_naissance"
+    },
+    {
+      "original": "CONDOM",
+      "replacement": "[LIEU_NAISS_1]",
+      "source": "regex",
+      "category": "lieu_naissance"
+    },
+    {
+      "original": "Buccalis",
+      "replacement": "[PERSONNE_1]",
+      "source": "ner",
+      "score": 0.9466555714607239
+    },
+    {
+      "original": "Madame [PATIENT_",
+      "replacement": "[PERSONNE_2]",
+      "source": "ner",
+      "score": 0.8676440119743347
+    },
+    {
+      "original": "Paul Jean",
+      "replacement": "[PERSONNE_3]",
+      "source": "ner",
+      "score": 0.8612924218177795
+    },
+    {
+      "original": "MATERNOWSKI",
+      "replacement": "[PERSONNE_4]",
+      "source": "ner",
+      "score": 0.9456329345703125
+    },
+    {
+      "original": "e) NAGOUAS",
+      "replacement": "[PERSONNE_5]",
+      "source": "ner",
+      "score": 0.8721564412117004
+    },
+    {
+      "original": "Klebsi",
+      "replacement": "[PERSONNE_6]",
+      "source": "ner",
+      "score": 0.8257390856742859
+    },
+    {
+      "original": "Citro",
+      "replacement": "[PERSONNE_7]",
+      "source": "ner",
+      "score": 0.8111836910247803
+    },
+    {
+      "original": "GILBERT Clara",
+      "replacement": "[PERSONNE_8]",
+      "source": "ner",
+      "score": 0.9892757534980774
+    },
+    {
+      "original": "[MEDECIN_22",
+      "replacement": "[PERSONNE_9]",
+      "source": "ner",
+      "score": 0.8607918620109558
+    },
+    {
+      "original": "[MEDECIN",
+      "replacement": "[PERSONNE_10]",
+      "source": "ner",
+      "score": 0.8795398473739624
+    },
+    {
+      "original": "Miction",
+      "replacement": "[PERSONNE_11]",
+      "source": "ner",
+      "score": 0.8751011490821838
+    },
+    {
+      "original": "Glyc",
+      "replacement": "[PERSONNE_12]",
+      "source": "ner",
+      "score": 0.9785528182983398
+    },
+    {
+      "original": "KUHN-RODRIGUEZ S",
+      "replacement": "[PERSONNE_13]",
+      "source": "ner",
+      "score": 0.9890475869178772
+    },
+    {
+      "original": "DALIA Amaïa",
+      "replacement": "[PERSONNE_14]",
+      "source": "ner",
+      "score": 0.9864678978919983
+    },
+    {
+      "original": "LEON DUQUE Juan david",
+      "replacement": "[PERSONNE_15]",
+      "source": "ner",
+      "score": 0.9987488389015198
+    },
+    {
+      "original": "DURANTEAU Ute Signé",
+      "replacement": "[PERSONNE_16]",
+      "source": "ner",
+      "score": 0.9973670840263367
+    },
+    {
+      "original": "MATERNOWSKI Paul Signé",
+      "replacement": "[PERSONNE_17]",
+      "source": "ner",
+      "score": 0.9965693950653076
+    },
+    {
+      "original": "TAGHVA PASSAND Goudarz",
+      "replacement": "[PERSONNE_18]",
+      "source": "ner",
+      "score": 0.9984901547431946
+    },
+    {
+      "original": "KUHN-RODRIGUEZ Marie-Pierre",
+      "replacement": "[PERSONNE_19]",
+      "source": "ner",
+      "score": 0.9985180497169495
+    },
+    {
+      "original": "CHEVALIER Lucile",
+      "replacement": "[PERSONNE_20]",
+      "source": "ner",
+      "score": 0.9965349435806274
+    },
+    {
+      "original": "Marie-Pierre",
+      "replacement": "[MEDECIN]",
+      "source": "ner",
+      "score": 0.939964771270752
+    },
+    {
+      "original": "NOGUES Sophie S",
+      "replacement": "[PERSONNE_21]",
+      "source": "ner",
+      "score": 0.8920299410820007
+    },
+    {
+      "original": "KUHN-RODRIGUEZ",
+      "replacement": "[PERSONNE]",
+      "source": "ner",
+      "score": 0.9982917904853821
+    },
+    {
+      "original": "Marie-Pierre Glyc",
+      "replacement": "[PERSONNE_22]",
+      "source": "ner",
+      "score": 0.9988271594047546
+    },
+    {
+      "original": "NOGUES Sophie",
+      "replacement": "[PERSONNE_23]",
+      "source": "ner",
+      "score": 0.956640362739563
+    },
+    {
+      "original": "Sophie S",
+      "replacement": "[PERSONNE_24]",
+      "source": "ner",
+      "score": 0.8866714239120483
+    },
+    {
+      "original": "LEON DUQUE Juan david S",
+      "replacement": "[PERSONNE_25]",
+      "source": "ner",
+      "score": 0.9372165203094482
+    },
+    {
+      "original": "CAZENAVE Xabi Quentin",
+      "replacement": "[PERSONNE_26]",
+      "source": "ner",
+      "score": 0.996477484703064
+    },
+    {
+      "original": "CAZENAVE Xabi Quentin Signé",
+      "replacement": "[PERSONNE_27]",
+      "source": "ner",
+      "score": 0.9971621632575989
+    },
+    {
+      "original": "Jean RECTALE",
+      "replacement": "[PERSONNE_28]",
+      "source": "ner",
+      "score": 0.9984656572341919
+    },
+    {
+      "original": "MATERNOWSKI Paul",
+      "replacement": "[PERSONNE_29]",
+      "source": "ner",
+      "score": 0.9986109137535095
+    },
+    {
+      "original": "MATERNOWSKI Paul S",
+      "replacement": "[PERSONNE_30]",
+      "source": "ner",
+      "score": 0.968824028968811
+    },
+    {
+      "original": "DURANTEAU Ute",
+      "replacement": "[PERSONNE_31]",
+      "source": "ner",
+      "score": 0.9962326884269714
+    },
+    {
+      "original": "CHEVALIER Lucile Signé",
+      "replacement": "[PERSONNE_32]",
+      "source": "ner",
+      "score": 0.9983345866203308
+    },
+    {
+      "original": "KUHN-RODRIGUEZ Marie-Pierre Signé",
+      "replacement": "[PERSONNE_33]",
+      "source": "ner",
+      "score": 0.9968046545982361
+    },
+    {
+      "original": "TAGHVA",
+      "replacement": "[PERSONNE]",
+      "source": "ner",
+      "score": 0.9975091218948364
+    },
+    {
+      "original": "Goudarz",
+      "replacement": "[MEDECIN]",
+      "source": "ner",
+      "score": 0.8684868812561035
+    },
+    {
+      "original": "SAND Goudarz",
+      "replacement": "[PERSONNE_34]",
+      "source": "ner",
+      "score": 0.9169244766235352
+    },
+    {
+      "original": "AND Goudarz",
+      "replacement": "[PERSONNE_35]",
+      "source": "ner",
+      "score": 0.942563533782959
+    },
+    {
+      "original": "TAGHVA ORALE",
+      "replacement": "[PERSONNE_36]",
+      "source": "ner",
+      "score": 0.9390201568603516
+    },
+    {
+      "original": "PASSAND Goudarz",
+      "replacement": "[PERSONNE_37]",
+      "source": "ner",
+      "score": 0.9787290692329407
+    },
+    {
+      "original": "Juan david",
+      "replacement": "[PERSONNE_38]",
+      "source": "ner",
+      "score": 0.9908421635627747
+    },
+    {
+      "original": "DUQUE Juan david",
+      "replacement": "[PERSONNE_39]",
+      "source": "ner",
+      "score": 0.9861922860145569
+    },
+    {
+      "original": "NOGUES Juan david",
+      "replacement": "[PERSONNE_40]",
+      "source": "ner",
+      "score": 0.9255251288414001
+    },
+    {
+      "original": "DUQUE",
+      "replacement": "[PERSONNE]",
+      "source": "ner",
+      "score": 0.8430215120315552
+    },
+    {
+      "original": "RODRIGUEZ Juan david",
+      "replacement": "[PERSONNE_41]",
+      "source": "ner",
+      "score": 0.9166897535324097
+    },
+    {
+      "original": "Sophie REDON",
+      "replacement": "[PERSONNE_42]",
+      "source": "ner",
+      "score": 0.9159219861030579
+    },
+    {
+      "original": "RODRIGUEZ",
+      "replacement": "[PERSONNE]",
+      "source": "ner",
+      "score": 0.9614688754081726
+    },
+    {
+      "original": "CAZENAVE Juan david",
+      "replacement": "[PERSONNE_43]",
+      "source": "ner",
+      "score": 0.8589810729026794
+    },
+    {
+      "original": "DUQUE DESINFECTION Juan david",
+      "replacement": "[PERSONNE_44]",
+      "source": "ner",
+      "score": 0.9569880366325378
+    },
+    {
+      "original": "Ines AMANN",
+      "replacement": "[SOIGNANT_11]",
+      "source": "ner",
+      "score": 0.9986129403114319
+    },
+    {
+      "original": "Pierre KUHN-",
+      "replacement": "[SOIGNANT_12]",
+      "source": "ner",
+      "score": 0.8906125426292419
+    },
+    {
+      "original": "a CHEVALIER",
+      "replacement": "[PERSONNE_45]",
+      "source": "ner",
+      "score": 0.8665874004364014
+    },
+    {
+      "original": "PASSAND",
+      "replacement": "[PERSONNE]",
+      "source": "ner",
+      "score": 0.9151932597160339
+    },
+    {
+      "original": "DURANTEAU",
+      "replacement": "[MEDECIN]",
+      "source": "ner",
+      "score": 0.9218727946281433
+    },
+    {
+      "original": "SIR",
+      "replacement": "[PERSONNE_46]",
+      "source": "ner",
+      "score": 0.8233469724655151
+    },
+    {
+      "original": "Goudarz ELIQUIS",
+      "replacement": "[PERSONNE_47]",
+      "source": "ner",
+      "score": 0.9950070977210999
+    },
+    {
+      "original": "AND Goudarz BACTRIM",
+      "replacement": "[PERSONNE_48]",
+      "source": "ner",
+      "score": 0.8932986855506897
+    },
+    {
+      "original": "Goudarz BACTRI",
+      "replacement": "[PERSONNE_49]",
+      "source": "ner",
+      "score": 0.8896738290786743
+    },
+    {
+      "original": "Pierre KUHN- POCHE",
+      "replacement": "[PERSONNE_50]",
+      "source": "ner",
+      "score": 0.8683809041976929
+    },
+    {
+      "original": "Pierre KUHN",
+      "replacement": "[PERSONNE_51]",
+      "source": "ner",
+      "score": 0.9980756044387817
+    },
+    {
+      "original": "Florence BIVES",
+      "replacement": "[PERSONNE_52]",
+      "source": "ner",
+      "score": 0.9926393628120422
+    },
+    {
+      "original": "Nolwenn BARRE",
+      "replacement": "[PERSONNE_53]",
+      "source": "ner",
+      "score": 0.9913205504417419
+    },
+    {
+      "original": "Kattina ARTOLA",
+      "replacement": "[PERSONNE_54]",
+      "source": "ner",
+      "score": 0.8843442797660828
+    },
+    {
+      "original": "GILBERT Clara Evolution",
+      "replacement": "[PERSONNE_55]",
+      "source": "ner",
+      "score": 0.972364604473114
+    },
+    {
+      "original": "Innohep",
+      "replacement": "[PERSONNE_56]",
+      "source": "ner",
+      "score": 0.9730066657066345
+    },
+    {
+      "original": "Lucile CHEVALIER",
+      "replacement": "[PERSONNE_57]",
+      "source": "ner",
+      "score": 0.9709216356277466
+    },
+    {
+      "original": "[MEDECIN_",
+      "replacement": "[PERSONNE_58]",
+      "source": "ner",
+      "score": 0.9839396476745605
+    },
+    {
+      "original": "DUCHET Serge FRANCE",
+      "replacement": "[PERSONNE_59]",
+      "source": "ner",
+      "score": 0.9872417449951172
+    },
+    {
+      "original": "GELOS LAURENT FRANCE",
+      "replacement": "[PERSONNE_60]",
+      "source": "ner",
+      "score": 0.9837981462478638
+    },
+    {
+      "original": "Salvador Allende",
+      "replacement": "[PERSONNE_61]",
+      "source": "ner",
+      "score": 0.9909833073616028
+    }
+  ]
+}