docslight-lite 0.1.0__py3-none-any.whl

docslight/static/app/docslight-extract.json

@@ -0,0 +1,307 @@
+{
+  "results": {
+    "name": {
+      "bboxes": [
+        {
+          "bbox": [
+            499,
+            84,
+            897,
+            122
+          ],
+          "page_id": 1
+        }
+      ],
+      "value": "合小合有限公司图书发货清单"
+    },
+    "仓储地址": {
+      "bboxes": [
+        {
+          "bbox": [
+            100,
+            405,
+            867,
+            472
+          ],
+          "page_id": 1
+        }
+      ],
+      "value": "广东佛山南海区前进东路与纵向五路交叉口博展物流园一期京东P库"
+    },
+    "制单日期": {
+      "bboxes": [
+        {
+          "bbox": [
+            87,
+            1091,
+            1054,
+            1161
+          ],
+          "page_id": 1
+        }
+      ],
+      "value": "2024-05-07"
+    },
+    "发货方式": {
+      "bboxes": [
+        {
+          "bbox": [
+            103,
+            162,
+            333,
+            262
+          ],
+          "page_id": 1
+        }
+      ],
+      "value": "汽运"
+    },
+    "审批日期": {
+      "bboxes": [
+        {
+          "bbox": [
+            103,
+            162,
+            333,
+            262
+          ],
+          "page_id": 1
+        }
+      ],
+      "value": "2024-05-07"
+    },
+    "审核日期": {
+      "bboxes": [
+        {
+          "bbox": [
+            87,
+            1091,
+            1054,
+            1161
+          ],
+          "page_id": 1
+        }
+      ],
+      "value": "2024-05-07"
+    },
+    "客户单号": {
+      "bboxes": [
+        {
+          "bbox": [
+            721,
+            158,
+            987,
+            221
+          ],
+          "page_id": 1
+        }
+      ],
+      "value": "5444412/1891133"
+    },
+    "开户银行": {
+      "bboxes": [
+        {
+          "bbox": [
+            99,
+            506,
+            623,
+            686
+          ],
+          "page_id": 1
+        }
+      ],
+      "value": "中国建设银行股份有限公司宿迁宿豫支行"
+    },
+    "总册数": {
+      "bboxes": [
+        {
+          "bbox": [
+            720,
+            265,
+            930,
+            398
+          ],
+          "page_id": 1
+        }
+      ],
+      "value": "250"
+    },
+    "批销单号": {
+      "bboxes": [
+        {
+          "bbox": [
+            721,
+            158,
+            987,
+            221
+          ],
+          "page_id": 1
+        }
+      ],
+      "value": "PXD222085"
+    },
+    "码洋": {
+      "bboxes": [
+        {
+          "bbox": [
+            720,
+            265,
+            930,
+            398
+          ],
+          "page_id": 1
+        }
+      ],
+      "value": "17469.00"
+    },
+    "移动电话": {
+      "bboxes": [
+        {
+          "bbox": [
+            99,
+            506,
+            623,
+            686
+          ],
+          "page_id": 1
+        }
+      ],
+      "value": "159700XXXX"
+    },
+    "税号": {
+      "bboxes": [
+        {
+          "bbox": [
+            99,
+            506,
+            623,
+            686
+          ],
+          "page_id": 1
+        }
+      ],
+      "value": "91321311562910"
+    },
+    "联系人": {
+      "bboxes": [
+        {
+          "bbox": [
+            99,
+            506,
+            623,
+            686
+          ],
+          "page_id": 1
+        }
+      ],
+      "value": "合小合"
+    },
+    "订单号": {
+      "bboxes": [
+        {
+          "bbox": [
+            103,
+            162,
+            333,
+            262
+          ],
+          "page_id": 1
+        }
+      ],
+      "value": "1750458"
+    },
+    "铁路到站": {
+      "bboxes": [
+        {
+          "bbox": [
+            720,
+            265,
+            930,
+            398
+          ],
+          "page_id": 1
+        }
+      ],
+      "value": "佛山市"
+    },
+    "tables": {
+      "Table_1": [
+        {
+          "ISBN": "978-7-5197-8886-5",
+          "包册数": "2+10(14）",
+          "单价": "49.00",
+          "图书名称": "“张三”身边的法律事 儿",
+          "实洋": "1,228.92",
+          "序号": "",
+          "折扣": "66.00",
+          "数量": "98",
+          "码洋": "才 862.00",
+          "货位号": "01-02-027-005"
+        },
+        {
+          "ISBN": "978-7-5197-9009-7",
+          "包册数": "0+3（8）",
+          "单价": "85.00",
+          "图书名称": "破产审判实务与前沿问 题研究",
+          "实洋": "168.30",
+          "序号": "2",
+          "折扣": "66.00",
+          "数量": "",
+          "码洋": "255.00",
+          "货位号": "01-02-063-002"
+        },
+        {
+          "ISBN": "978-7-5197-8939-8",
+          "包册数": "11+2(5）",
+          "单价": "108.00",
+          "图书名称": "中华人民共和国公司法 及司法解释指导莱例全 书",
+          "实洋": "4,062.96",
+          "序号": "3",
+          "折扣": "",
+          "数量": "57",
+          "码洋": "",
+          "货位号": "01-04-020-004"
+        },
+        {
+          "ISBN": "978-7-5197-8953-4",
+          "包册数": "11+2（5）",
+          "单价": "108.00",
+          "图书名称": "中华人民共和国民法典 及司法解释指导案例全 书",
+          "实洋": "4,062.96",
+          "序号": "4",
+          "折扣": "",
+          "数量": "57",
+          "码洋": "",
+          "货位号": "01-04-029-006"
+        },
+        {
+          "ISBN": "978-7-5197-9057-8",
+          "包册数": "7+4(13)",
+          "单价": "32.00",
+          "图书名称": "民事起诉状、答辩状示 范文本（试行）",
+          "实洋": "1,216.00",
+          "序号": "1CD",
+          "折扣": "40.00",
+          "数量": "95",
+          "码洋": "3,040.00",
+          "货位号": "01-04-288-005"
+        }
+      ]
+    },
+    "_table_bboxes": {
+      "Table_1": {
+        "bbox": [
+          85,
+          694,
+          1342,
+          1026
+        ],
+        "page_id": 1
+      }
+    }
+  },
+  "metadata": {
+    "source_width": 1425,
+    "source_height": 1233
+  }
+}

docslight/static/app/extract.js

@@ -0,0 +1,394 @@
+import {
+  bindDropzone,
+  bindResultTabs,
+  downloadText,
+  highlightBboxes,
+  initHealthBadge,
+  loadPreview,
+  normalizeExtractPayload,
+  postForm,
+  renderExtractCards,
+  renderJsonView,
+  renderPlaceholder,
+  renderPreview,
+  setFormError,
+} from "./common.js";
+import { initI18n, onLanguageChange, t } from "./i18n.js";
+
+const extractForm = document.querySelector("#extractForm");
+const modeSelect = document.querySelector("#modeSelect");
+const cloudConfig = document.querySelector("#cloudConfig");
+const cloudExtractMode = document.querySelector("#cloudExtractMode");
+const groundingToggle = document.querySelector("#groundingToggle");
+const localLlmBlock = document.querySelector("#localLlmBlock");
+const fileInput = document.querySelector("#fileInput");
+const dropZone = document.querySelector("#dropZone");
+const fileName = document.querySelector("#fileName");
+const previewTitle = document.querySelector("#previewTitle");
+const previewCanvas = document.querySelector("#previewCanvas");
+const officePreviewNotice = document.querySelector("#officePreviewNotice");
+const highlightStatus = document.querySelector("#highlightStatus");
+const formError = document.querySelector("#formError");
+const submitButton = document.querySelector("#submitButton");
+const downloadButton = document.querySelector("#downloadButton");
+const metadataPreview = document.querySelector("#metadataPreview");
+const fieldTemplateName = document.querySelector("#fieldTemplateName");
+const fieldsRows = document.querySelector("#fieldsRows");
+const addFieldButton = document.querySelector("#addFieldButton");
+const addTableButton = document.querySelector("#addTableButton");
+const fieldsPanel = document.querySelector("#fieldsPanel");
+const jsonPanel = document.querySelector("#jsonPanel");
+const extractResultTabs = document.querySelector("#extractResultTabs");
+const healthStatus = document.querySelector("#healthStatus");
+
+const state = {
+  fieldRows: [],
+  currentTab: "fields",
+  hasResult: false,
+  latestJson: "",
+  resultSource: "cloud",
+  previewRequestId: 0,
+};
+
+function syncRuntimeControls() {
+  const isCloud = modeSelect?.value !== "local";
+  if (cloudConfig) cloudConfig.hidden = !isCloud;
+  if (localLlmBlock) localLlmBlock.hidden = isCloud;
+  if (groundingToggle) groundingToggle.hidden = !isCloud || cloudExtractMode?.value !== "integrate";
+}
+
+function normalizedMeta(row) {
+  const prompt = row.prompt?.trim() || null;
+  const mapping = row.mapping?.trim() || null;
+  return { prompt, mapping };
+}
+
+function nextTableNumber() {
+  const numbers = state.fieldRows
+    .filter((row) => row.type === "table")
+    .map((row) => Number(row.name.replace(/^Table_/, "")))
+    .filter(Number.isFinite);
+  return numbers.length ? Math.max(...numbers) + 1 : 1;
+}
+
+function inputControl(value, placeholder, onInput) {
+  const input = document.createElement("input");
+  input.type = "text";
+  input.value = value || "";
+  input.placeholder = placeholder;
+  input.addEventListener("input", () => onInput(input.value));
+  return input;
+}
+
+function labeledCell(labelText, control) {
+  const label = document.createElement("label");
+  label.className = "field-label";
+  const span = document.createElement("span");
+  span.textContent = labelText;
+  label.append(span, control);
+  return label;
+}
+
+function actionButton(text, onClick) {
+  const button = document.createElement("button");
+  button.type = "button";
+  button.className = "ghost-button";
+  button.textContent = text;
+  button.addEventListener("click", onClick);
+  return button;
+}
+
+const PROMPT_ICON_SVG = '<svg viewBox="0 0 16 16" fill="none" stroke="currentColor" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round"><path d="M11.5 1.5l3 3L5 14H2v-3L11.5 1.5z"/></svg>';
+
+function closeAllPopovers() {
+  document.querySelectorAll(".prompt-popover-backdrop").forEach((el) => el.remove());
+  document.querySelectorAll(".prompt-popover").forEach((el) => el.remove());
+}
+
+function nameWithPromptCell(row, labelText, namePlaceholder, updateName, updatePrompt) {
+  const wrapper = document.createElement("div");
+  wrapper.className = "prompt-cell";
+
+  const nameInput = inputControl(row.name, namePlaceholder, updateName);
+  const label = document.createElement("label");
+  label.className = "field-label";
+  const labelSpan = document.createElement("span");
+  labelSpan.textContent = labelText;
+  label.append(labelSpan, nameInput);
+
+  const trigger = document.createElement("button");
+  trigger.type = "button";
+  trigger.className = "prompt-trigger";
+  trigger.setAttribute("aria-label", t("fields.prompt"));
+  trigger.innerHTML = PROMPT_ICON_SVG;
+  if (row.prompt?.trim()) trigger.dataset.active = "true";
+
+  trigger.addEventListener("click", (event) => {
+    event.stopPropagation();
+    if (document.body.contains(document.querySelector(".prompt-popover"))) {
+      closeAllPopovers();
+      return;
+    }
+    closeAllPopovers();
+
+    const backdrop = document.createElement("div");
+    backdrop.className = "prompt-popover-backdrop";
+    backdrop.addEventListener("click", closeAllPopovers);
+
+    const popover = document.createElement("div");
+    popover.className = "prompt-popover";
+    popover.addEventListener("click", (e) => e.stopPropagation());
+    popover.addEventListener("mousedown", (e) => e.stopPropagation());
+
+    const header = document.createElement("div");
+    header.className = "prompt-popover-header";
+    header.textContent = t("fields.prompt");
+    popover.append(header);
+
+    const textarea = document.createElement("textarea");
+    textarea.value = row.prompt || "";
+    textarea.placeholder = t("fields.promptPlaceholder");
+    textarea.addEventListener("input", () => {
+      row.prompt = textarea.value;
+      trigger.dataset.active = textarea.value.trim() ? "true" : "false";
+    });
+    popover.append(textarea);
+
+    const rect = trigger.getBoundingClientRect();
+    popover.style.top = `${rect.bottom + 8}px`;
+    popover.style.left = `${Math.max(8, rect.right - 260)}px`;
+
+    document.body.append(backdrop, popover);
+    textarea.focus();
+  });
+
+  wrapper.append(label, trigger);
+  return wrapper;
+}
+
+function renderTextRow(row) {
+  const card = document.createElement("article");
+  card.className = "field-row-card";
+  const title = document.createElement("strong");
+  title.textContent = t("fields.field");
+  card.append(
+    title,
+    nameWithPromptCell(row, t("fields.name"), t("fields.namePlaceholder"), (value) => { row.name = value; }, (value) => { row.prompt = value; }),
+    labeledCell(t("fields.mapping"), inputControl(row.mapping, t("fields.mappingPlaceholder"), (value) => { row.mapping = value; })),
+    actionButton(t("fields.remove"), () => {
+      state.fieldRows = state.fieldRows.filter((candidate) => candidate !== row);
+      renderFieldsBuilder();
+    }),
+  );
+  return card;
+}
+
+function renderColumnRow(table, column) {
+  const row = document.createElement("div");
+  row.className = "field-row-card";
+  row.append(
+    nameWithPromptCell(column, t("fields.column"), t("fields.columnNamePlaceholder"), (value) => { column.name = value; }, (value) => { column.prompt = value; }),
+    labeledCell(t("fields.mapping"), inputControl(column.mapping, t("fields.mappingPlaceholder"), (value) => { column.mapping = value; })),
+    actionButton(t("fields.removeColumn"), () => {
+      table.columns = table.columns.filter((candidate) => candidate !== column);
+      renderFieldsBuilder();
+    }),
+  );
+  return row;
+}
+
+function renderTableRow(row) {
+  const card = document.createElement("article");
+  card.className = "table-field-card";
+  const title = document.createElement("strong");
+  title.textContent = t("fields.table");
+  const columns = document.createElement("div");
+  columns.className = "table-columns";
+  row.columns.forEach((column) => columns.append(renderColumnRow(row, column)));
+  card.append(
+    title,
+    labeledCell(t("fields.tableName"), inputControl(row.name, t("fields.tableNamePlaceholder"), (value) => { row.name = value; })),
+    columns,
+    actionButton(t("fields.addColumn"), () => {
+      row.columns.push({ name: t("fields.defaultColumnName"), prompt: "", mapping: "" });
+      renderFieldsBuilder();
+    }),
+    actionButton(t("fields.removeTable"), () => {
+      state.fieldRows = state.fieldRows.filter((candidate) => candidate !== row);
+      renderFieldsBuilder();
+    }),
+  );
+  return card;
+}
+
+function renderFieldsBuilder() {
+  if (!fieldsRows) return;
+  fieldsRows.replaceChildren();
+  state.fieldRows.forEach((row) => {
+    fieldsRows.append(row.type === "table" ? renderTableRow(row) : renderTextRow(row));
+  });
+}
+
+function addTextField(name = t("fields.initialFieldName")) {
+  state.fieldRows.push({ type: "field", name, prompt: "", mapping: "" });
+  renderFieldsBuilder();
+}
+
+function addTableField(name = `Table_${nextTableNumber()}`, columnName = t("fields.initialColumnName")) {
+  state.fieldRows.push({
+    type: "table",
+    name,
+    columns: [{ name: columnName, prompt: "", mapping: "" }],
+  });
+  renderFieldsBuilder();
+}
+
+function buildFieldsPayload() {
+  const payload = {
+    keys: {},
+    tableHeaders: {},
+    name: fieldTemplateName?.value.trim() || "Document",
+  };
+
+  state.fieldRows.forEach((row) => {
+    if (row.type === "field") {
+      const name = row.name.trim();
+      if (name) payload.keys[name] = normalizedMeta(row);
+      return;
+    }
+
+    const tableName = row.name.trim();
+    const columns = {};
+    row.columns.forEach((column) => {
+      const columnName = column.name.trim();
+      if (columnName) columns[columnName] = normalizedMeta(column);
+    });
+    if (tableName && Object.keys(columns).length) payload.tableHeaders[tableName] = columns;
+  });
+
+  if (!Object.keys(payload.keys).length) delete payload.keys;
+  if (!Object.keys(payload.tableHeaders).length) delete payload.tableHeaders;
+  return payload;
+}
+
+function refreshPreview() {
+  return loadPreview({
+    fileInput,
+    previewTitle,
+    previewCanvas,
+    officePreviewNotice,
+    highlightStatus,
+    state,
+  });
+}
+
+function renderEmptyResult() {
+  const placeholder = t("extract.placeholder");
+  renderPlaceholder(fieldsPanel, placeholder);
+  renderPlaceholder(jsonPanel, placeholder);
+  state.hasResult = false;
+  state.latestJson = "";
+  if (metadataPreview) metadataPreview.textContent = t("extract.metadataEmpty");
+  if (downloadButton) downloadButton.disabled = true;
+}
+
+function renderExtractResult(result) {
+  state.hasResult = true;
+  const normalized = normalizeExtractPayload(result);
+  const displayData = normalized.results;
+  state.latestJson = JSON.stringify(normalized.full, null, 2);
+  state.resultSource = modeSelect?.value === "local" ? "local" : "cloud";
+
+  renderExtractCards(displayData, fieldsPanel, {
+    source: state.resultSource,
+    onPick: (boxes, source) => {
+      if (!boxes) {
+        highlightBboxes(null, source || state.resultSource, { previewCanvas, highlightStatus });
+        return;
+      }
+      highlightBboxes(boxes, source || state.resultSource, { previewCanvas, highlightStatus });
+    },
+  });
+  renderJsonView(normalized.full, jsonPanel);
+
+  if (metadataPreview) {
+    metadataPreview.textContent = JSON.stringify(normalized.metadata, null, 2);
+  }
+  if (downloadButton) downloadButton.disabled = false;
+}
+
+function validateForm(fields) {
+  if (!fileInput?.files?.length) return t("error.selectDocument");
+
+  const isCloud = modeSelect?.value !== "local";
+  const apiKey = extractForm?.querySelector('[name="api_key"]')?.value?.trim();
+  if (isCloud && !apiKey) return t("error.cloudApiKeyRequired");
+
+  const localProvider = extractForm?.querySelector('[name="local_llm_provider"]')?.value?.trim();
+  const localModel = extractForm?.querySelector('[name="local_llm_model"]')?.value?.trim();
+  const localBaseUrl = extractForm?.querySelector('[name="local_llm_base_url"]')?.value?.trim();
+  if (!isCloud && (!localProvider || !localModel || !localBaseUrl)) {
+    return t("error.localLlmRequired");
+  }
+
+  if (!fields.keys && !fields.tableHeaders) return t("error.fieldsRequired");
+  return "";
+}
+
+function refreshLocalizedDynamicCopy() {
+  renderFieldsBuilder();
+  if (!state.hasResult) renderEmptyResult();
+  renderPreview(state.preview || null, { previewTitle, previewCanvas, officePreviewNotice, highlightStatus, state });
+}
+
+modeSelect?.addEventListener("change", syncRuntimeControls);
+cloudExtractMode?.addEventListener("change", syncRuntimeControls);
+addFieldButton?.addEventListener("click", () => addTextField(t("fields.defaultFieldName")));
+addTableButton?.addEventListener("click", () => addTableField());
+
+extractForm?.addEventListener("submit", async (event) => {
+  event.preventDefault();
+  setFormError(formError, "");
+
+  const fields = buildFieldsPayload();
+  const validationError = validateForm(fields);
+  if (validationError) {
+    setFormError(formError, validationError);
+    return;
+  }
+
+  const body = new FormData(extractForm);
+  body.set("file", fileInput.files[0]);
+  body.set("fields", JSON.stringify(fields));
+  if (modeSelect?.value !== "local") {
+    body.set("cloud_extract_mode", cloudExtractMode?.value || "vlm");
+    if (cloudExtractMode?.value !== "integrate") body.delete("enable_grounding");
+  }
+  if (submitButton) submitButton.disabled = true;
+
+  try {
+    const payload = await postForm("/api/extract", body);
+    renderExtractResult(payload.result || payload);
+  } catch (error) {
+    setFormError(formError, error instanceof Error ? error.message : t("extract.failed"));
+  } finally {
+    if (submitButton) submitButton.disabled = false;
+  }
+});
+
+downloadButton?.addEventListener("click", () => {
+  downloadText(state.latestJson, "docslight-extract.json");
+});
+
+initI18n();
+initHealthBadge(healthStatus);
+bindDropzone({ dropZone, fileInput, fileName, onFileChange: refreshPreview });
+bindResultTabs(extractResultTabs, (tab) => {
+  state.currentTab = tab;
+});
+onLanguageChange(refreshLocalizedDynamicCopy);
+syncRuntimeControls();
+renderPreview(null, { previewTitle, previewCanvas, officePreviewNotice, highlightStatus, state });
+renderEmptyResult();
+addTextField(t("fields.initialFieldName"));
+addTableField("Table_1", t("fields.initialColumnName"));