refactor: move arquivos para buildar lib

Author: andrewalkermo

.github/workflows/extrai_fatos.yml

@@ -37,7 +37,7 @@ jobs:
               echo "[JSON] input=$filename flags=[$flags]"
               # remove stale json output if any
               rm -f ./outputs/output.json
-              docker run --rm -v "$(pwd)":/dptoie_python dptoie_python poetry run python3 src/main.py -i "$input_path" $input_args $flags || echo "json-run-failed for $filename $flags"
+              docker run --rm -v "$(pwd)":/dptoie_python dptoie_python poetry run python3 src/dptoie_python/main.py -i "$input_path" $input_args $flags || echo "json-run-failed for $filename $flags"
 
               # compute cleaned suffix for filename (empty => base name only)
               if [[ -z "$flags" ]]; then
@@ -58,7 +58,7 @@ jobs:
               echo "[CSV] input=$filename flags=[$flags]"
               # remove stale csv output if any
               rm -f ./outputs/output.csv
-              docker run --rm -v "$(pwd)":/dptoie_python dptoie_python poetry run python3 src/main.py -i "$input_path" $input_args $flags -o ./outputs/output.csv -ot csv || echo "csv-run-failed for $filename $flags"
+              docker run --rm -v "$(pwd)":/dptoie_python dptoie_python poetry run python3 src/dptoie_python/main.py -i "$input_path" $input_args $flags -o ./outputs/output.csv -ot csv || echo "csv-run-failed for $filename $flags"
 
               if [[ -z "$flags" ]]; then
                 out_csv="./outputs/${name}.csv"

.gitignore

@@ -6,3 +6,4 @@ out.txt
 !/outputs/.gitkeep
 /models/*
 !/models/.gitkeep
+/dist/

README-ptbr.md

@@ -3,7 +3,7 @@
 
 Extrator de Informação Aberta para língua portuguesa baseado em análise de dependências (SpaCy + Stanza).
 
-Este guia mostra todas as formas de rodar o projeto pelo `src/main.py`, com todas as variações de argumentos, tanto localmente (Poetry) quanto com Docker/Docker Compose.
+Este guia mostra todas as formas de rodar o projeto pelo `src/dptoie_python/main.py`, com todas as variações de argumentos, tanto localmente (Poetry) quanto com Docker/Docker Compose.
 
 - Requisitos mínimos: Python 3.12+, Poetry, ou Docker (opcional)
 - Modelos: o Stanza faz o download automático na primeira execução. Você pode definir `STANZA_RESOURCES_DIR` para usar um diretório local de modelos (ex.: `./models/.stanza_resources`).
@@ -30,7 +30,7 @@ poetry install
 Forma geral:
 
 ```bash
-poetry run python3 src/main.py \
+poetry run python3 src/dptoie_python/main.py \
   -i <caminho_entrada> \
   -it <txt|conll> \
   -o <caminho_saida> \
@@ -60,37 +60,37 @@ Importante:
 
 1) TXT de entrada, JSON de saída (padrões):
     ```bash
-    poetry run python3 src/main.py -i ./inputs/ceten-200.txt -it txt -o ./outputs/out.json -ot json
+    poetry run python3 src/dptoie_python/main.py -i ./inputs/ceten-200.txt -it txt -o ./outputs/out.json -ot json
     ```
 
 2) TXT de entrada, CSV de saída, ativando coordenação e sujeito oculto:
     ```bash
-    poetry run python3 src/main.py -i ./inputs/ceten-200.txt -it txt -o ./outputs/out.csv -ot csv -cc
+    poetry run python3 src/dptoie_python/main.py -i ./inputs/ceten-200.txt -it txt -o ./outputs/out.csv -ot csv -cc
     ```
 
 3) TXT de entrada, saída em texto legível:
     ```bash
-    poetry run python3 src/main.py -i ./inputs/ceten-200.txt -it txt -o ./outputs/out.txt -ot txt -cc -sc -a -t
+    poetry run python3 src/dptoie_python/main.py -i ./inputs/ceten-200.txt -it txt -o ./outputs/out.txt -ot txt -cc -sc -a -t
     ```
 
 4) Entrada já em CoNLL-U, JSON de saída:
     ```bash
-    poetry run python3 src/main.py -i ./inputs/teste.conll -it conll -o ./outputs/out.json -ot json -cc -sc -a -t
+    poetry run python3 src/dptoie_python/main.py -i ./inputs/teste.conll -it conll -o ./outputs/out.json -ot json -cc -sc -a -t
     ```
 
 5) Somente conjunções coordenativas:
     ```bash
-    poetry run python3 src/main.py -i ./inputs/ceten-200.txt -it txt -o ./outputs/cc.json -ot json -cc
+    poetry run python3 src/dptoie_python/main.py -i ./inputs/ceten-200.txt -it txt -o ./outputs/cc.json -ot json -cc
     ```
 
 6) Modo debug para inspeção detalhada:
     ```bash
-    poetry run python3 src/main.py -i ./inputs/ceten-200.txt -it txt -o ./outputs/out.json -ot json -cc -debug
+    poetry run python3 src/dptoie_python/main.py -i ./inputs/ceten-200.txt -it txt -o ./outputs/out.json -ot json -cc -debug
     ```
 
 7) Ver lista de argumentos:
     ```bash
-    poetry run python3 src/main.py -h
+    poetry run python3 src/dptoie_python/main.py -h
     ```
 
 Saídas esperadas:
@@ -112,7 +112,7 @@ docker run --rm -it \
   -v "$(pwd)":/dptoie_python \
   -w /dptoie_python \
   dptoie_python \
-  poetry run python3 src/main.py -i /dptoie_python/inputs/teste.conll -it conll -o /dptoie_python/outputs/out.json -ot json -cc -sc -a -t
+  poetry run python3 src/dptoie_python/main.py -i /dptoie_python/inputs/teste.conll -it conll -o /dptoie_python/outputs/out.json -ot json -cc -sc -a -t
 ```
 
 Observação: ajuste os caminhos de `-i` e `-o` conforme necessário; use `-it txt` quando a entrada for texto linha-a-linha.
@@ -122,7 +122,7 @@ Observação: ajuste os caminhos de `-i` e `-o` conforme necessário; use `-it t
 O arquivo `docker-compose.yml` já inclui o serviço `dptoie_python`. Você pode editar a linha `command:` para o cenário desejado. Exemplo de comando recomendado:
 
 ```yaml
-command: poetry run python3 src/main.py -i /dptoie_python/inputs/teste.conll -it conll -o /dptoie_python/outputs/out.json -ot json -cc -sc -a -t
+command: poetry run python3 src/dptoie_python/main.py -i /dptoie_python/inputs/teste.conll -it conll -o /dptoie_python/outputs/out.json -ot json -cc -sc -a -t
 ```
 
 Então rode:
@@ -132,7 +132,7 @@ docker compose up --build
 
 Use o comando `run` para executar outros comandos personalizados:
 ```bash
-docker compose run dptoie_python poetry run python3 src/main.py -i /dptoie_python/inputs/ceten-200.txt -it txt -o /dptoie_python/outputs/out.csv -ot csv -cc
+docker compose run dptoie_python poetry run python3 src/dptoie_python/main.py -i /dptoie_python/inputs/ceten-200.txt -it txt -o /dptoie_python/outputs/out.csv -ot csv -cc
 ```
 
 

README.md

@@ -2,7 +2,7 @@
 
 Open Information Extractor for Portuguese based on dependency analysis (SpaCy + Stanza).
 
-This guide shows all ways to run the project via `src/main.py`, with all argument variations, both locally (Poetry) and with Docker / Docker Compose.
+This guide shows all ways to run the project via `src/dptoie_python/main.py`, with all argument variations, both locally (Poetry) and with Docker / Docker Compose.
 
 - Minimum requirements: Python 3.12+, Poetry, or Docker (optional)
 - Models: Stanza downloads models automatically on first run. You can set `STANZA_RESOURCES_DIR` to use a local models directory (e.g., `./models/.stanza_resources`).
@@ -29,7 +29,7 @@ poetry install
 General form:
 
 ```bash
-poetry run python3 src/main.py \
+poetry run python3 src/dptoie_python/main.py \
   -i <input_path> \
   -it <txt|conll> \
   -o <output_path> \
@@ -59,37 +59,37 @@ Important:
 
 1) TXT input, JSON output (defaults):
 ```bash
-poetry run python3 src/main.py -i ./inputs/ceten-200.txt -it txt -o ./outputs/out.json -ot json
+poetry run python3 src/dptoie_python/main.py -i ./inputs/ceten-200.txt -it txt -o ./outputs/out.json -ot json
 ```
 
 2) TXT input, CSV output, enabling coordination and hidden subject (flag example):
 ```bash
-poetry run python3 src/main.py -i ./inputs/ceten-200.txt -it txt -o ./outputs/out.csv -ot csv -cc
+poetry run python3 src/dptoie_python/main.py -i ./inputs/ceten-200.txt -it txt -o ./outputs/out.csv -ot csv -cc
 ```
 
 3) TXT input, human-readable text output:
 ```bash
-poetry run python3 src/main.py -i ./inputs/ceten-200.txt -it txt -o ./outputs/out.txt -ot txt -cc -sc -a -t
+poetry run python3 src/dptoie_python/main.py -i ./inputs/ceten-200.txt -it txt -o ./outputs/out.txt -ot txt -cc -sc -a -t
 ```
 
 4) Input already in CoNLL-U, JSON output:
 ```bash
-poetry run python3 src/main.py -i ./inputs/teste.conll -it conll -o ./outputs/out.json -ot json -cc -sc -a -t
+poetry run python3 src/dptoie_python/main.py -i ./inputs/teste.conll -it conll -o ./outputs/out.json -ot json -cc -sc -a -t
 ```
 
 5) Only coordinating conjunctions:
 ```bash
-poetry run python3 src/main.py -i ./inputs/ceten-200.txt -it txt -o ./outputs/cc.json -ot json -cc
+poetry run python3 src/dptoie_python/main.py -i ./inputs/ceten-200.txt -it txt -o ./outputs/cc.json -ot json -cc
 ```
 
 6) Debug mode for detailed inspection:
 ```bash
-poetry run python3 src/main.py -i ./inputs/ceten-200.txt -it txt -o ./outputs/out.json -ot json -cc -debug
+poetry run python3 src/dptoie_python/main.py -i ./inputs/ceten-200.txt -it txt -o ./outputs/out.json -ot json -cc -debug
 ```
 
 7) Show arguments list:
 ```bash
-poetry run python3 src/main.py -h
+poetry run python3 src/dptoie_python/main.py -h
 ```
 
 Expected outputs:
@@ -111,7 +111,7 @@ docker run --rm -it \
   -v "$(pwd)":/dptoie_python \
   -w /dptoie_python \
   dptoie_python \
-  poetry run python3 src/main.py -i /dptoie_python/inputs/teste.conll -it conll -o /dptoie_python/outputs/out.json -ot json -cc -sc -a -t
+  poetry run python3 src/dptoie_python/main.py -i /dptoie_python/inputs/teste.conll -it conll -o /dptoie_python/outputs/out.json -ot json -cc -sc -a -t
 ```
 
 Note: adjust the `-i` and `-o` paths as needed; use `-it txt` when the input is line-by-line text.
@@ -121,7 +121,7 @@ Note: adjust the `-i` and `-o` paths as needed; use `-it txt` when the input is
 The `docker-compose.yml` file already includes the `dptoie_python` service. You can edit the `command:` line for the desired scenario. Example recommended command:
 
 ```yaml
-command: poetry run python3 src/main.py -i /dptoie_python/inputs/teste.conll -it conll -o /dptoie_python/outputs/out.json -ot json -cc -sc -a -t
+command: poetry run python3 src/dptoie_python/main.py -i /dptoie_python/inputs/teste.conll -it conll -o /dptoie_python/outputs/out.json -ot json -cc -sc -a -t
 ```
 
 Then run:
@@ -131,7 +131,7 @@ docker compose up --build
 
 Use `run` to execute other custom commands:
 ```bash
-docker compose run dptoie_python poetry run python3 src/main.py -i /dptoie_python/inputs/ceten-200.txt -it txt -o /dptoie_python/outputs/out.csv -ot csv -cc
+docker compose run dptoie_python poetry run python3 src/dptoie_python/main.py -i /dptoie_python/inputs/ceten-200.txt -it txt -o /dptoie_python/outputs/out.csv -ot csv -cc
 ```
 
 Tips:

docker-compose.yml

@@ -6,6 +6,6 @@ services:
     container_name: dptoie_python
     volumes:
       - .:/dptoie_python
-    command: poetry run python3 src/main.py -i /dptoie_python/inputs/teste.conll -it conll -o /dptoie_python/outputs/output.json -ot json -cc -sc -hs -a -t
+    command: poetry run python3 src/dptoie_python/main.py -i /dptoie_python/inputs/teste.conll -it conll -o /dptoie_python/outputs/output.json -ot json -cc -sc -hs -a -t
     environment:
       - STANZA_RESOURCES_DIR=/dptoie_python/models/.stanza_resources

pyproject.toml

@@ -4,7 +4,7 @@ version = "0.1.0"
 description = "DPToie-Python is an Open Information Extractor for Portuguese language that employs Dependence Parser and Part of Speech Tagger models with Stanford CoreNLP."
 authors = ["André Walker <[email protected]>", "Rafael Glauber <[email protected]>", "Daniela Barreiro Claro <[email protected]>"]
 readme = "README.md"
-package-mode = false
+package-mode = true
 
 [tool.poetry.dependencies]
 python = ">=3.12,<3.14"

src/__init__.py renamed to src/dptoie_python/__init__.py

@@ -14,7 +14,7 @@
 from spacy import Language
 from typing import Any, Generator
 from spacy_conll.parser import ConllParser
-from src.extraction import Extractor, ExtractorConfig, Extraction
+from src.dptoie_python.extraction import Extractor, ExtractorConfig, Extraction
 
 
 def generate_conll_file_from_sentences_file(input_file: str) -> str: