yescent/par
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix: JSON无页面规则时兜底展示原始结构，清理重复代码

Browse Source
This commit is contained in:
mediabot-pt
2026-07-01 10:20:40 +08:00
parent df6928d0cb
commit 1dd600c713
1 changed files with 67 additions and 52 deletions
Download Patch File Download Diff File Expand all files Collapse all files

119
par-core/src/main/java/com/par/core/adapter/AdapterEngine.java
View File

@@ -41,62 +41,77 @@ public class AdapterEngine {
result.put("contentType", "json");
Map.Entry<String, PageConfig> matched = matchPage(config.getPages(), accessUrl);
if (matched == null) {
result.put("matched", false);
result.put("message", "未匹配到页面规则");
result.put("availablePages", new ArrayList<>(config.getPages() != null ? config.getPages().keySet() : Set.of()));
return result;
}
PageConfig page = matched.getValue();
result.put("matched", true);
result.put("pageName", matched.getKey());
if (matched != null && matched.getValue().getFields() != null && !matched.getValue().getFields().isEmpty()) {
// 有页面规则 → 按 selector 提取
PageConfig page = matched.getValue();
result.put("matched", true);
result.put("pageName", matched.getKey());
if (page.getFields() == null || page.getFields().isEmpty()) {
result.put("message", "页面规则未定义字段");
result.put("extracted", List.of());
result.put("rowCount", 0);
result.put("fieldCount", 0);
return result;
}
// 列表容器
List<Object> items = new ArrayList<>();
if (page.getListContainer() != null && !page.getListContainer().isBlank()) {
Object container = resolveJsonPath(parsed, page.getListContainer());
if (container instanceof List) items = (List<Object>) container;
else items.add(parsed);
} else {
items.add(parsed);
}
// 逐行提取字段
List<Map<String, Object>> rows = new ArrayList<>();
int max = Math.min(items.size(), 50);
for (int i = 0; i < max; i++) {
Map<String, Object> row = new LinkedHashMap<>();
boolean hasData = false;
for (Map.Entry<String, FieldExtractor> field : page.getFields().entrySet()) {
String fieldName = field.getKey();
FieldExtractor fe = field.getValue();
if (fe.getSelector() == null) continue;
Object val = resolveJsonPath(items.get(i), fe.getSelector());
if (val != null) hasData = true;
if (fe.getOptional() != null && fe.getOptional() && val == null) continue;
if (val instanceof String s) {
row.put(fieldName, applyTransform(s, fe.getTransform()));
} else if (val != null) {
row.put(fieldName, toDisplayString(val));
} else {
row.put(fieldName, null);
}
List<Object> items = new ArrayList<>();
if (page.getListContainer() != null && !page.getListContainer().isBlank()) {
Object container = resolveJsonPath(parsed, page.getListContainer());
if (container instanceof List) items = (List<Object>) container;
else items.add(parsed);
} else {
items.add(parsed);
}
if (hasData) rows.add(row);
List<Map<String, Object>> rows = new ArrayList<>();
int max = Math.min(items.size(), 50);
for (int i = 0; i < max; i++) {
Map<String, Object> row = new LinkedHashMap<>();
boolean hasData = false;
for (Map.Entry<String, FieldExtractor> field : page.getFields().entrySet()) {
String fieldName = field.getKey();
FieldExtractor fe = field.getValue();
if (fe.getSelector() == null) continue;
Object val = resolveJsonPath(items.get(i), fe.getSelector());
if (val != null) hasData = true;
if (fe.getOptional() != null && fe.getOptional() && val == null) continue;
if (val instanceof String s) {
row.put(fieldName, applyTransform(s, fe.getTransform()));
} else if (val != null) {
row.put(fieldName, toDisplayString(val));
} else {
row.put(fieldName, null);
}
}
if (hasData) rows.add(row);
}
result.put("extracted", rows);
result.put("rowCount", rows.size());
result.put("fieldCount", page.getFields().size());
result.put("message", rows.isEmpty() ? "selector 未匹配到数据" : "共 " + rows.size() + " 行");
} else {
// 无页面规则 → 兜底展示原始结构
String reason = matched == null ? "未匹配到页面规则" : "页面规则未定义字段";
result.put("matched", true);
result.put("pageName", "JSON原始结构");
result.put("message", reason + "，展示原始数据");
List<Map<String, Object>> rows = new ArrayList<>();
if (parsed instanceof Map) {
Map<String, Object> row = new LinkedHashMap<>();
for (String key : ((Map<String, Object>) parsed).keySet()) {
row.put(key, toDisplayString(((Map<String, Object>) parsed).get(key)));
}
rows.add(row);
} else if (parsed instanceof List) {
int m = Math.min(((List<?>) parsed).size(), 10);
for (int i = 0; i < m; i++) {
Map<String, Object> row = new LinkedHashMap<>();
row.put("value", ((List<?>) parsed).get(i));
rows.add(row);
}
} else {
Map<String, Object> row = new LinkedHashMap<>();
row.put("value", parsed);
rows.add(row);
}
result.put("extracted", rows);
result.put("rowCount", rows.size());
result.put("fieldCount", rows.isEmpty() ? 0 : rows.get(0).size());
}
result.put("extracted", rows);
result.put("rowCount", rows.size());
result.put("fieldCount", page.getFields().size());
result.put("message", rows.isEmpty() ? "selector 未匹配到数据" : "共 " + rows.size() + " 行");
} catch (Exception e) {
result.put("contentType", "json");
result.put("matched", false);