@mdgf11/filesystem-lib 2.2.11 → 2.2.13

package/dist/filesystemDocumentMethods.d.ts

@@ -1,7 +1,7 @@
 import { PartialJurisprudenciaDocument } from "@stjiris/jurisprudencia-document";
-import { ContentType, Date_Area_Section, FilesystemDocument, Retrievable_Metadata, Sharepoint_Metadata } from "./types.js";
+import { ContentType, FilesystemDocument } from "./types.js";
 export declare function writeFilesystemDocument(filesystem_document: FilesystemDocument): void;
 export declare function loadFilesystemDocument(jsonPath: string): FilesystemDocument;
-export declare function createJurisprudenciaDocument(retrievable_Metadata: Retrievable_Metadata, contents: ContentType[], date_area_section: Date_Area_Section, sharepoint_metadata?: Sharepoint_Metadata): Promise<PartialJurisprudenciaDocument>;
 export declare function hasSelectableText(buffer: Buffer): Promise<boolean>;
 export declare function generateFilePath(jurisprudencia_document: PartialJurisprudenciaDocument): string;
+export declare function extractContent(contents: ContentType[]): Promise<string[]>;

package/dist/filesystemDocumentMethods.js

@@ -1,8 +1,6 @@
-import { calculateHASH, calculateUUID } from "@stjiris/jurisprudencia-document";
 import fs from "fs";
 import mammoth from "mammoth";
 import { DETAILS_NAME, FILESYSTEM_PATH, ORIGINAL_NAME, ROOT_PATH, SHAREPOINT_COPY_PATH } from "./types.js";
-import { DescritorOficial } from "./descritores.js";
 import { getDocument } from "pdfjs-dist/legacy/build/pdf.mjs";
 export function writeFilesystemDocument(filesystem_document) {
     if (!filesystem_document.content)
@@ -56,77 +54,6 @@ export function loadFilesystemDocument(jsonPath) {
         }))
     };
 }
-export async function createJurisprudenciaDocument(retrievable_Metadata, contents, date_area_section, sharepoint_metadata) {
-    if (!retrievable_Metadata) {
-        throw new Error("Missing metadata.");
-    }
-    const content = await extractContent(contents);
-    const url = sharepoint_metadata ? sharepoint_metadata.sharepoint_url : "";
-    let Original = {};
-    let CONTENT = content;
-    let numProc = retrievable_Metadata.process_number;
-    let Data = Intl.DateTimeFormat("pt-PT").format(date_area_section.file_date);
-    let origin = "STJ (Sharepoint)";
-    Original["Decisão Texto Integral"] = content.map(line => `<p><font>${line}</font><br>`).join('');
-    Original["Data"] = Data;
-    Original["Número de Processo"] = numProc;
-    Original["Fonte"] = origin;
-    Original["URL"] = url;
-    Original["Jurisprudência"] = "Simples";
-    let obj = {
-        "Original": Original,
-        "CONTENT": CONTENT,
-        "Data": Data,
-        "Número de Processo": numProc,
-        "Fonte": origin,
-        "URL": url,
-        "Jurisprudência": { Index: ["Simples"], Original: ["Simples"], Show: ["Simples"] },
-        "STATE": "importação",
-    };
-    if (retrievable_Metadata.process_mean.includes("Sumário")) {
-        obj.Sumário = content.map(line => `<p><font>${line}</font><br>`).join('');
-        obj.Texto = "";
-    }
-    else {
-        obj.Sumário = "";
-        obj.Texto = content.map(line => `<p><font>${line}</font><br>`).join('');
-    }
-    if (retrievable_Metadata.descriptors && retrievable_Metadata.descriptors.length > 0) {
-        obj.Descritores = {
-            Index: retrievable_Metadata.descriptors.map(desc => DescritorOficial[desc]),
-            Original: retrievable_Metadata.descriptors,
-            Show: retrievable_Metadata.descriptors.map(desc => DescritorOficial[desc])
-        };
-    }
-    if (date_area_section.area && date_area_section.area.length > 0) {
-        obj.Área = { Index: [date_area_section.area], Original: [date_area_section.area], Show: [date_area_section.area] };
-    }
-    if (date_area_section.section && date_area_section.section.length > 0) {
-        obj.Secção = { Index: [date_area_section.section], Original: [date_area_section.section], Show: [date_area_section.section] };
-    }
-    if (retrievable_Metadata.judge && retrievable_Metadata.judge.length > 0) {
-        obj["Relator Nome Profissional"] = { Index: [retrievable_Metadata.judge], Original: [retrievable_Metadata.judge], Show: [retrievable_Metadata.judge] };
-    }
-    if (retrievable_Metadata.process_mean && retrievable_Metadata.process_mean.length > 0) {
-        obj["Meio Processual"] = { Index: retrievable_Metadata.process_mean, Original: retrievable_Metadata.process_mean, Show: retrievable_Metadata.process_mean };
-    }
-    if (retrievable_Metadata.decision && retrievable_Metadata.decision.length > 0) {
-        obj["Decisão"] = { Index: [retrievable_Metadata.decision], Original: [retrievable_Metadata.decision], Show: [retrievable_Metadata.decision] };
-    }
-    obj["HASH"] = calculateHASH({
-        ...obj,
-        Original: obj.Original,
-        "Número de Processo": obj["Número de Processo"],
-        Data: obj.Data,
-        "Meio Processual": obj["Meio Processual"],
-        "Texto": obj.Texto,
-        "Texto Não Anonimizado": obj.Texto,
-        "Sumário": obj.Sumário,
-        "Sumário Não Anonimizado": obj.Sumário,
-    });
-    obj["UUID"] = calculateUUID(obj["HASH"]);
-    return obj;
-}
 export async function hasSelectableText(buffer) {
     try {
         const uint8Array = new Uint8Array(buffer);
@@ -157,14 +84,7 @@ export function generateFilePath(jurisprudencia_document) {
     const data = parseDateFromString(jurisprudencia_document.Data);
     return `/${jurisprudencia_document.Área.Show}/${data.getFullYear()}/${data.getMonth() + 1}/${data.getDate()}/${jurisprudencia_document["Número de Processo"].replace("/", "-")}/${jurisprudencia_document.UUID}`;
 }
-function parseDateFromString(dateStr) {
-    const [dayStr, monthStr, yearStr] = dateStr.split("/");
-    const day = parseInt(dayStr, 10);
-    const month = parseInt(monthStr, 10) - 1;
-    const year = yearStr ? parseInt(yearStr, 10) : new Date().getFullYear();
-    return new Date(year, month, day);
-}
-async function extractContent(contents) {
+export async function extractContent(contents) {
     for (const content of contents) {
         if (content.extension === "txt") {
             return content.data.toString('utf-8').split(/\r?\n/).filter(line => line.trim().length > 0);
@@ -178,6 +98,13 @@ async function extractContent(contents) {
     }
     throw new Error("Contents are not a supported format.");
 }
+function parseDateFromString(dateStr) {
+    const [dayStr, monthStr, yearStr] = dateStr.split("/");
+    const day = parseInt(dayStr, 10);
+    const month = parseInt(monthStr, 10) - 1;
+    const year = yearStr ? parseInt(yearStr, 10) : new Date().getFullYear();
+    return new Date(year, month, day);
+}
 async function pdfToLines(buffer) {
     const uint8Array = new Uint8Array(buffer);
     const loadingTask = getDocument({ data: uint8Array, verbosity: 0 });

package/dist/types.d.ts

@@ -48,7 +48,7 @@ export type Date_Area_Section = {
     area: string;
     section: string;
 };
-export declare const SUPPORTED_EXTENSIONS: readonly ["txt", "pdf", "docx", "html"];
+export declare const SUPPORTED_EXTENSIONS: readonly ["txt", "pdf", "docx", "html", "json"];
 export type Supported_Content_Extensions = typeof SUPPORTED_EXTENSIONS[number];
 export type ContentType = {
     extension: Supported_Content_Extensions;

package/dist/types.js

@@ -12,7 +12,7 @@ export const DGSI_LOGS_PATH = `${LOGS_PATH}/DGSI`;
 export const SHAREPOINT_UPDATE_DIR = `${ROOT_PATH}${SHAREPOINT_LOGS_PATH}`;
 export const DGSI_UPDATE_DIR = `${ROOT_PATH}${DGSI_LOGS_PATH}`;
 export const SOURCE_TO_PATH = { "STJ (Sharepoint)": SHAREPOINT_UPDATE_DIR, "DGSI": DGSI_UPDATE_DIR };
-export const SUPPORTED_EXTENSIONS = ["txt", "pdf", "docx", "html"];
+export const SUPPORTED_EXTENSIONS = ["txt", "pdf", "docx", "html", "json"];
 export function isSupportedExtension(ext) {
     return SUPPORTED_EXTENSIONS.includes(ext);
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@mdgf11/filesystem-lib",
-  "version": "2.2.11",
+  "version": "2.2.13",
   "description": "Library to extend usage of jurisprudencia-document",
   "license": "ISC",
   "author": "Miguel Fonseca",

package/src/filesystemDocumentMethods.ts

@@ -1,4 +1,4 @@
-import { calculateHASH, calculateUUID, JurisprudenciaDocument, PartialJurisprudenciaDocument } from "@stjiris/jurisprudencia-document";
+import { PartialJurisprudenciaDocument } from "@stjiris/jurisprudencia-document";
 import fs from "fs";
 import mammoth from "mammoth";
 import { ContentType, Date_Area_Section, DETAILS_NAME, FILESYSTEM_PATH, FilesystemDocument, ORIGINAL_NAME, Retrievable_Metadata, ROOT_PATH, SHAREPOINT_COPY_PATH, Sharepoint_Metadata, SupportedUpdateSources } from "./types.js";
@@ -63,82 +63,6 @@ export function loadFilesystemDocument(jsonPath: string): FilesystemDocument {
     };
 }
 
-export async function createJurisprudenciaDocument(retrievable_Metadata: Retrievable_Metadata, contents: ContentType[], date_area_section: Date_Area_Section, sharepoint_metadata?: Sharepoint_Metadata): Promise<PartialJurisprudenciaDocument> {
-    if (!retrievable_Metadata) {
-        throw new Error("Missing metadata.");
-    }
-    const content = await extractContent(contents);
-    const url = sharepoint_metadata ? sharepoint_metadata.sharepoint_url : "";
-
-    let Original: JurisprudenciaDocument["Original"] = {};
-    let CONTENT: JurisprudenciaDocument["CONTENT"] = content;
-    let numProc: JurisprudenciaDocument["Número de Processo"] = retrievable_Metadata.process_number;
-    let Data: JurisprudenciaDocument["Data"] = Intl.DateTimeFormat("pt-PT").format(date_area_section.file_date);
-    let origin: SupportedUpdateSources = "STJ (Sharepoint)";
-
-    Original["Decisão Texto Integral"] = content.map(line => `<p><font>${line}</font><br>`).join('');
-    Original["Data"] = Data;
-    Original["Número de Processo"] = numProc;
-    Original["Fonte"] = origin;
-    Original["URL"] = url;
-    Original["Jurisprudência"] = "Simples";
-
-    let obj: PartialJurisprudenciaDocument = {
-        "Original": Original,
-        "CONTENT": CONTENT,
-        "Data": Data,
-        "Número de Processo": numProc,
-        "Fonte": origin,
-        "URL": url,
-        "Jurisprudência": { Index: ["Simples"], Original: ["Simples"], Show: ["Simples"] },
-        "STATE": "importação",
-    }
-    if (retrievable_Metadata.process_mean.includes("Sumário")) {
-        obj.Sumário = content.map(line => `<p><font>${line}</font><br>`).join('');
-        obj.Texto = "";
-    } else {
-        obj.Sumário = "";
-        obj.Texto = content.map(line => `<p><font>${line}</font><br>`).join('');
-    }
-    if (retrievable_Metadata.descriptors && retrievable_Metadata.descriptors.length > 0) {
-        obj.Descritores = {
-            Index: retrievable_Metadata.descriptors.map(desc => DescritorOficial[desc]),
-            Original: retrievable_Metadata.descriptors,
-            Show: retrievable_Metadata.descriptors.map(desc => DescritorOficial[desc])
-        }
-    }
-    if (date_area_section.area && date_area_section.area.length > 0) {
-        obj.Área = { Index: [date_area_section.area], Original: [date_area_section.area], Show: [date_area_section.area] };
-    }
-    if (date_area_section.section && date_area_section.section.length > 0) {
-        obj.Secção = { Index: [date_area_section.section], Original: [date_area_section.section], Show: [date_area_section.section] };
-    }
-    if (retrievable_Metadata.judge && retrievable_Metadata.judge.length > 0) {
-        obj["Relator Nome Profissional"] = { Index: [retrievable_Metadata.judge], Original: [retrievable_Metadata.judge], Show: [retrievable_Metadata.judge] };
-    }
-    if (retrievable_Metadata.process_mean && retrievable_Metadata.process_mean.length > 0) {
-        obj["Meio Processual"] = { Index: retrievable_Metadata.process_mean, Original: retrievable_Metadata.process_mean, Show: retrievable_Metadata.process_mean };
-    }
-    if (retrievable_Metadata.decision && retrievable_Metadata.decision.length > 0) {
-        obj["Decisão"] = { Index: [retrievable_Metadata.decision], Original: [retrievable_Metadata.decision], Show: [retrievable_Metadata.decision] };
-    }
-
-    obj["HASH"] = calculateHASH({
-        ...obj,
-        Original: obj.Original,
-        "Número de Processo": obj["Número de Processo"],
-        Data: obj.Data,
-        "Meio Processual": obj["Meio Processual"],
-        "Texto": obj.Texto,
-        "Texto Não Anonimizado": obj.Texto,
-        "Sumário": obj.Sumário,
-        "Sumário Não Anonimizado": obj.Sumário,
-    })
-
-    obj["UUID"] = calculateUUID(obj["HASH"]);
-    return obj;
-}
-
 export async function hasSelectableText(buffer: Buffer): Promise<boolean> {
     try {
         const uint8Array = new Uint8Array(buffer);
@@ -175,17 +99,7 @@ export function generateFilePath(jurisprudencia_document: PartialJurisprudenciaD
     return `/${jurisprudencia_document.Área.Show}/${data.getFullYear()}/${data.getMonth() + 1}/${data.getDate()}/${jurisprudencia_document["Número de Processo"].replace("/", "-")}/${jurisprudencia_document.UUID}`
 }
 
-function parseDateFromString(dateStr: string): Date {
-    const [dayStr, monthStr, yearStr] = dateStr.split("/");
-
-    const day = parseInt(dayStr, 10);
-    const month = parseInt(monthStr, 10) - 1;
-    const year = yearStr ? parseInt(yearStr, 10) : new Date().getFullYear();
-
-    return new Date(year, month, day);
-}
-
-async function extractContent(contents: ContentType[]): Promise<string[]> {
+export async function extractContent(contents: ContentType[]): Promise<string[]> {
     for (const content of contents) {
         if (content.extension === "txt") {
             return content.data.toString('utf-8').split(/\r?\n/).filter(line => line.trim().length > 0);
@@ -200,6 +114,16 @@ async function extractContent(contents: ContentType[]): Promise<string[]> {
     throw new Error("Contents are not a supported format.");
 }
 
+function parseDateFromString(dateStr: string): Date {
+    const [dayStr, monthStr, yearStr] = dateStr.split("/");
+
+    const day = parseInt(dayStr, 10);
+    const month = parseInt(monthStr, 10) - 1;
+    const year = yearStr ? parseInt(yearStr, 10) : new Date().getFullYear();
+
+    return new Date(year, month, day);
+}
+
 async function pdfToLines(buffer: Buffer): Promise<string[]> {
     const uint8Array = new Uint8Array(buffer);
     const loadingTask = getDocument({ data: uint8Array, verbosity: 0 });

package/src/types.ts

@@ -1,7 +1,6 @@
 
 export const UpdateSources = ["STJ (Sharepoint)", "DGSI"] as const;
 export type SupportedUpdateSources = typeof UpdateSources[number];
-import { PartialJurisprudenciaDocument } from '@stjiris/jurisprudencia-document';
 import dotenv from 'dotenv';
 
 dotenv.config();
@@ -46,7 +45,7 @@ export type Sharepoint_Metadata = {
 export type Retrievable_Metadata = { process_number: string, judge: string, process_mean: string[], decision: string, descriptors?: string[] };
 export type Date_Area_Section = { file_date: Date, area: string, section: string };
 
-export const SUPPORTED_EXTENSIONS = ["txt", "pdf", "docx", "html"] as const;
+export const SUPPORTED_EXTENSIONS = ["txt", "pdf", "docx", "html", "json"] as const;
 export type Supported_Content_Extensions = typeof SUPPORTED_EXTENSIONS[number];
 
 export type ContentType = {