工具类，案件压缩包导入优化

2023-12-12 16:22:41 +08:00
parent 5730bb12a7
commit a78e3e23d6
28 changed files with 1876 additions and 1096 deletions
@@ -118,10 +118,10 @@ public class OCRUtils {
        for (FatchRule fatchRule : fatchRules) {
            // 从后往前抓取
            if (fatchRule.getFatchOrder() != null && fatchRule.getFatchOrder() == 1) {
-                reverseSubstringText(ocrText, fatchRule, fatchMap);
+                reverseSubstringText(ocrText, fatchRule, fatchMap,null);
            } else {
                // 从前往后抓取
-                substringText(ocrText, fatchRule, fatchMap);
+                substringText(ocrText, fatchRule, fatchMap,null);

            }
        }
@@ -129,43 +129,32 @@ public class OCRUtils {

    }

-    public static void sub1(String text, FatchRule fatchRule, Map<String, String> fatchMap) {
-        if (StrUtil.isEmpty(fatchRule.getStartContent()) && StrUtil.isEmpty(fatchRule.getEndContent())) {
-            fatchMap.put(fatchRule.getColumnName(), trimStr(text));
-        } else if (StrUtil.isNotEmpty(fatchRule.getStartContent())) {
-            int startContIndex = StrUtil.ordinalIndexOf(text, fatchRule.getStartContent(), fatchRule.getStartContentRepeatOrder());
-            if (startContIndex != -1) {
-                // 开始不为空结束为空
-                if (StrUtil.isEmpty(fatchRule.getEndContent())) {
-                    if ((startContIndex + fatchRule.getStartContent().length()) <= text.length()) {
-                        String substring = text.substring(startContIndex + fatchRule.getStartContent().length());
-                        // 去除\n
-                        fatchMap.put(fatchRule.getColumnName(), trimStr(substring));
-                    }
+    /**
+     * 根据抓取规则获取内容
+     *
+     * @param ocrText    ocr识别的text
+     * @param fatchRules 抓取规则
+     * @return
+     */
+    public static void fatchRuleGetContent(String ocrText, List<FatchRule> fatchRules, Map<String, String> fatchMap, Long caseId) {
+        if (StrUtil.isEmpty(ocrText) || CollectionUtil.isEmpty(fatchRules)) {
+            return;
+        }
+        for (FatchRule fatchRule : fatchRules) {
+            // 从后往前抓取
+            if (fatchRule.getFatchOrder() != null && fatchRule.getFatchOrder() == 1) {
+                reverseSubstringText(ocrText, fatchRule, fatchMap, caseId);
+            } else {
+                // 从前往后抓取
+                substringText(ocrText, fatchRule, fatchMap, caseId);

-                } else {
-                    // 开始不为空结束不为空
-                    int endContIndex = StrUtil.ordinalIndexOf(text, fatchRule.getEndContent(), fatchRule.getEndContentRepeatOrder());
-                    if (endContIndex != -1 && endContIndex <= text.length() && (startContIndex + fatchRule.getStartContent().length()) <= endContIndex) {
-                        String substring = text.substring(startContIndex + fatchRule.getStartContent().length(), endContIndex);
-                        // 去除\n
-                        fatchMap.put(fatchRule.getColumnName(), trimStr(substring));
-                    }
-
-                }
-//
-            }
-
-        } else if (StrUtil.isEmpty(fatchRule.getStartContent()) && StrUtil.isNotEmpty(fatchRule.getEndContent())) {
-            // 开始为空结束不为空
-            int endContIndex = StrUtil.ordinalIndexOf(text, fatchRule.getEndContent(), fatchRule.getEndContentRepeatOrder());
-            if (endContIndex != -1) {
-                String substring = text.substring(0, endContIndex);
-                fatchMap.put(fatchRule.getColumnName(), trimStr(substring));
            }
        }
+
+
    }

+
    /**
     * 正向截取字段
     *
@@ -173,19 +162,19 @@ public class OCRUtils {
     * @param fatchRule
     * @param fatchMap
     */
-    private static void substringText(String text, FatchRule fatchRule, Map<String, String> fatchMap) {
+    private static void substringText(String text, FatchRule fatchRule, Map<String, String> fatchMap, Long caseId) {
        String startContent = fatchRule.getStartContent();
        String endContent = fatchRule.getEndContent();
        // 开始为空结束为空
        if (StrUtil.isEmpty(startContent) && StrUtil.isEmpty(endContent)) {
-            fatchMap.put(fatchRule.getColumnName(), trimStr(text));
+            fatchMap.put(fatchRule.getColumnName()+Constants.PDFSTR+caseId, trimStr(text));
        } else if (StrUtil.isNotEmpty(startContent) && StrUtil.isEmpty(endContent)) {
            // 开始不为空结束为空
            int startContIndex = StrUtil.ordinalIndexOf(text, startContent, fatchRule.getStartContentRepeatOrder());
            if (startContIndex != -1 && text.length() >= (startContIndex + startContent.length())) {
                String substring = text.substring(startContIndex + startContent.length());
                // 去除\n
-                fatchMap.put(fatchRule.getColumnName(), trimStr(substring));
+                fatchMap.put(fatchRule.getColumnName()+Constants.PDFSTR+caseId, trimStr(substring));
            }

        } else if (StrUtil.isEmpty(startContent) && StrUtil.isNotEmpty(endContent)) {
@@ -193,7 +182,7 @@ public class OCRUtils {
            int endContIndex = StrUtil.ordinalIndexOf(text, endContent, fatchRule.getEndContentRepeatOrder());
            if (endContIndex != -1) {
                String substring = text.substring(0, endContIndex);
-                fatchMap.put(fatchRule.getColumnName(), trimStr(substring));
+                fatchMap.put(fatchRule.getColumnName()+Constants.PDFSTR+caseId, trimStr(substring));
            }
        } else if (StrUtil.isNotEmpty(startContent) && StrUtil.isNotEmpty(endContent)) {
            // 开始结束不为空
@@ -202,12 +191,72 @@ public class OCRUtils {
            if (startIndexOf != -1 && endIndexOf != -1 && endIndexOf >= (startIndexOf + startContent.length()) && text.length() >= endIndexOf) {
                String substring = text.substring(startIndexOf + startContent.length(), endIndexOf);

-                fatchMap.put(fatchRule.getColumnName(), trimStr(substring));
+                fatchMap.put(fatchRule.getColumnName()+Constants.PDFSTR+caseId, trimStr(substring));

            }
        }
    }

+
+    /**
+     * 从后往前截取字符串
+     *
+     * @param text
+     * @param fatchRule
+     * @param fatchMap
+     */
+    public static void reverseSubstringText(String text, FatchRule fatchRule, Map<String, String> fatchMap, Long caseId) {
+
+        // 反正字符串
+        String reverseText = StrUtil.reverse(text);
+        // 结束截取字段
+        String reverseEndContent = "";
+        // 开始截取字段
+        String reverseStartContent = "";
+        if (StrUtil.isNotEmpty(fatchRule.getEndContent())) {
+            reverseEndContent = StrUtil.reverse(fatchRule.getEndContent());
+        }
+        if (StrUtil.isNotEmpty(fatchRule.getStartContent())) {
+            reverseStartContent = StrUtil.reverse(fatchRule.getStartContent());
+        }
+        // 开始和结束截取都为空
+        if (StrUtil.isEmpty(reverseStartContent) && StrUtil.isEmpty(reverseEndContent)) {
+            fatchMap.put(fatchRule.getColumnName()+Constants.PDFSTR+caseId, trimStr(text));
+        } else if (StrUtil.isEmpty(reverseStartContent) && StrUtil.isNotEmpty(reverseEndContent)) {
+            // 开始为空，结束不为空
+            // 根据截取的序号查找出位置
+            int indexOf = StrUtil.ordinalIndexOf(reverseText, reverseEndContent, fatchRule.getEndContentRepeatOrder());
+            if (indexOf != -1) {
+                String substring = reverseText.substring(0, indexOf);
+                if (StrUtil.isNotEmpty(substring)) {
+                    fatchMap.put(fatchRule.getColumnName()+Constants.PDFSTR+caseId, trimStr(StrUtil.reverse(substring)));
+                }
+            }
+        } else if (StrUtil.isNotEmpty(reverseStartContent) && StrUtil.isEmpty(reverseEndContent)) {
+            // 开始不为空，结束为空
+            int indexOf = StrUtil.ordinalIndexOf(reverseText, reverseStartContent, fatchRule.getStartContentRepeatOrder());
+            if (indexOf != -1 && (indexOf + reverseStartContent.length() <= text.length())) {
+                String substring = reverseText.substring(indexOf + reverseStartContent.length());
+                if (StrUtil.isNotEmpty(substring)) {
+                    fatchMap.put(fatchRule.getColumnName()+Constants.PDFSTR+caseId, trimStr(StrUtil.reverse(substring)));
+                }
+            }
+
+        } else if (StrUtil.isNotEmpty(reverseStartContent) && StrUtil.isNotEmpty(reverseEndContent)) {
+            // 开始结束都不为空
+            int endIndexOf = StrUtil.ordinalIndexOf(reverseText, reverseEndContent, fatchRule.getEndContentRepeatOrder());
+            int startIndexOf = StrUtil.ordinalIndexOf(reverseText, reverseStartContent, fatchRule.getStartContentRepeatOrder());
+            if (startIndexOf != -1 && endIndexOf != -1 && endIndexOf >= (startIndexOf + reverseStartContent.length()) && text.length() >= endIndexOf) {
+                String substring = reverseText.substring(startIndexOf + reverseStartContent.length(), endIndexOf);
+                if (StrUtil.isNotEmpty(substring)) {
+                    fatchMap.put(fatchRule.getColumnName()+Constants.PDFSTR+caseId, trimStr(StrUtil.reverse(substring)));
+                }
+            }
+        }
+
+
+    }
+
    /**
     * 去除末尾空格
     *
@@ -223,63 +272,4 @@ public class OCRUtils {

    }

-    /**
-     * 从后往前截取字符串
-     *
-     * @param text
-     * @param fatchRule
-     * @param fatchMap
-     */
-    public static void reverseSubstringText(String text, FatchRule fatchRule, Map<String, String> fatchMap) {
-
-        // 反正字符串
-        String reverseText = StrUtil.reverse(text);
-        // 结束截取字段
-        String reverseEndContent = "";
-        // 开始截取字段
-        String reverseStartContent = "";
-        if (StrUtil.isNotEmpty(fatchRule.getEndContent())) {
-            reverseEndContent = StrUtil.reverse(fatchRule.getEndContent());
-        }
-        if (StrUtil.isNotEmpty(fatchRule.getStartContent())) {
-            reverseStartContent = StrUtil.reverse(fatchRule.getStartContent());
-        }
-        // 开始和结束截取都为空
-        if (StrUtil.isEmpty(reverseStartContent) && StrUtil.isEmpty(reverseEndContent)) {
-            fatchMap.put(fatchRule.getColumnName(), trimStr(text));
-        } else if (StrUtil.isEmpty(reverseStartContent) && StrUtil.isNotEmpty(reverseEndContent)) {
-            // 开始为空，结束不为空
-            // 根据截取的序号查找出位置
-            int indexOf = StrUtil.ordinalIndexOf(reverseText, reverseEndContent, fatchRule.getEndContentRepeatOrder());
-            if (indexOf != -1) {
-                String substring = reverseText.substring(0, indexOf);
-                if (StrUtil.isNotEmpty(substring)) {
-                    fatchMap.put(fatchRule.getColumnName(), trimStr(StrUtil.reverse(substring)));
-                }
-            }
-        } else if (StrUtil.isNotEmpty(reverseStartContent) && StrUtil.isEmpty(reverseEndContent)) {
-            // 开始不为空，结束为空
-            int indexOf = StrUtil.ordinalIndexOf(reverseText, reverseStartContent, fatchRule.getStartContentRepeatOrder());
-            if (indexOf != -1 && (indexOf + reverseStartContent.length() <= text.length())) {
-                String substring = reverseText.substring(indexOf + reverseStartContent.length());
-                if (StrUtil.isNotEmpty(substring)) {
-                    fatchMap.put(fatchRule.getColumnName(), trimStr(StrUtil.reverse(substring)));
-                }
-            }
-
-        } else if (StrUtil.isNotEmpty(reverseStartContent) && StrUtil.isNotEmpty(reverseEndContent)) {
-            // 开始结束都不为空
-            int endIndexOf = StrUtil.ordinalIndexOf(reverseText, reverseEndContent, fatchRule.getEndContentRepeatOrder());
-            int startIndexOf = StrUtil.ordinalIndexOf(reverseText, reverseStartContent, fatchRule.getStartContentRepeatOrder());
-            if (startIndexOf != -1 && endIndexOf != -1 && endIndexOf >= (startIndexOf + reverseStartContent.length()) && text.length() >= endIndexOf) {
-                String substring = reverseText.substring(startIndexOf + reverseStartContent.length(), endIndexOf);
-                if (StrUtil.isNotEmpty(substring)) {
-                    fatchMap.put(fatchRule.getColumnName(), trimStr(StrUtil.reverse(substring)));
-                }
-            }
-        }
-
-
-    }
-
 }