fix: 完善数据预处理与 JSON 上传链路
This commit is contained in:
caoxiaozhu
@@ -183,8 +183,36 @@ for (const field of ['sourceStart', 'sourceEnd', 'originalContent', 'editedConte
|
||||
assert.ok(typesSource.includes(field), `PreviewItem 缺少字段：${field}`)
|
||||
}
|
||||
assert.match(typesSource, /sourceFileId/, 'PreviewItem 缺少来源文件标识')
|
||||
assert.match(typesSource, /sourceLocator\?: PreviewSourceLocator/, 'PreviewItem 缺少结构化来源定位契约')
|
||||
assert.match(typesSource, /headingPath\?: string\[\]/, 'PreviewItem 缺少非结构化标题路径')
|
||||
assert.match(typesSource, /PreviewSourceLocatorKind = 'json' \| 'jsonl' \| 'csv' \| 'xlsx'/, '前端来源定位 kind 未使用明确联合类型')
|
||||
assert.match(contractTypesSource, /DataProcessSourceLocatorKind = 'json' \| 'jsonl' \| 'csv' \| 'xlsx'/, 'API 来源定位 kind 未使用明确联合类型')
|
||||
for (const field of ['kind', 'record_index', 'start_line', 'end_line', 'source_start', 'source_end', 'json_pointer', 'sheet_index', 'sheet_name', 'row_number', 'sheet_record_index']) {
|
||||
assert.ok(typesSource.includes(field), `PreviewSourceLocator 缺少字段：${field}`)
|
||||
assert.ok(contractTypesSource.includes(field), `后端来源定位契约缺少字段：${field}`)
|
||||
}
|
||||
assert.match(contractTypesSource, /source_locator\?: DataProcessSourceLocator/, '质量信息缺少来源定位契约')
|
||||
assert.match(contractTypesSource, /heading_path\?: string\[\]/, '质量信息缺少标题路径契约')
|
||||
assert.match(viewSource, /const sourceLocator = item\.quality_score\?\.source_locator/, '预览映射丢失来源定位')
|
||||
assert.match(viewSource, /sourceStart:\s*item\.source_start\s*\?\?\s*sourceLocator\?\.source_start/, 'JSON locator 的字符起点没有映射到预览项')
|
||||
assert.match(viewSource, /sourceEnd:\s*item\.source_end\s*\?\?\s*sourceLocator\?\.source_end/, 'JSON locator 的字符终点没有映射到预览项')
|
||||
assert.match(viewSource, /sourceStartLine:\s*item\.source_start_line\s*\?\?\s*sourceLocator\?\.start_line/, 'JSON locator 的起始行没有映射到预览项')
|
||||
assert.match(viewSource, /sourceEndLine:\s*item\.source_end_line\s*\?\?\s*sourceLocator\?\.end_line/, 'JSON locator 的结束行没有映射到预览项')
|
||||
assert.match(viewSource, /headingPath:[\s\S]*?item\.quality_score\?\.heading_path/, '预览映射丢失标题路径')
|
||||
assert.match(typesSource, /export type StepId = 'create' \| 'model' \| 'upload' \| 'preview' \| 'generate' \| 'results'/, '步骤类型缺少独立大模型选择步骤')
|
||||
assert.match(modelSource, /export function sourceLines/, '缺少源文件行偏移生成函数')
|
||||
assert.match(modelSource, /export function sourceLineWindow/, '缺少有界源文件行窗口函数')
|
||||
assert.match(modelSource, /maxLines:\s*number/, '源文件行窗口缺少最大渲染行数参数')
|
||||
assert.doesNotMatch(modelSource, /\.split\(\s*['"]\\n['"]\s*\)/, '源文件行窗口仍会先对全文 split')
|
||||
assert.match(modelSource, /lines\.length < limit/, '源文件行扫描没有受最大行数约束')
|
||||
assert.match(modelSource, /unicodeCodePointLength/, '源文件字符偏移未与后端 Unicode code point 计数保持一致')
|
||||
assert.match(modelSource, /export function sourceLineNumberAtOffset/, '字符偏移缺少无数组的行号解析函数')
|
||||
const manualPreviewHelperStart = modelSource.indexOf('export function isManualPreviewItem(')
|
||||
const manualPreviewHelperEnd = modelSource.indexOf('\n}', manualPreviewHelperStart)
|
||||
assert.ok(manualPreviewHelperStart >= 0, '缺少统一的手动预览项判定函数')
|
||||
const manualPreviewHelperSource = modelSource.slice(manualPreviewHelperStart, manualPreviewHelperEnd + 2)
|
||||
for (const field of ['status', 'originalContent', 'sourceStart', 'sourceEnd', 'sourceStartLine', 'sourceEndLine', 'sourcePages', 'sourceLocator']) {
|
||||
assert.ok(manualPreviewHelperSource.includes(field), `手动预览项判定缺少来源字段：${field}`)
|
||||
}
|
||||
assert.doesNotMatch(modelSource, /buildPreviewItems/, '前端不应保留与后端重复的本地切片算法')
|
||||
assert.match(viewSource, /selectedPreviewFileId/, '父页面缺少当前预览文件状态')
|
||||
const previewBuildBindingStart = viewSource.indexOf('useDataProcessPreviewBuild()')
|
||||
@@ -210,6 +238,30 @@ for (const marker of [
|
||||
}
|
||||
assert.match(previewSource, /sourceStart/, '第四步未使用来源起始偏移')
|
||||
assert.match(previewSource, /sourceEnd/, '第四步未使用来源结束偏移')
|
||||
const lineRangeStart = previewSource.indexOf('function lineRange(item: PreviewItem)')
|
||||
const lineRangeEnd = previewSource.indexOf('\n}', lineRangeStart)
|
||||
const lineRangeSource = previewSource.slice(lineRangeStart, lineRangeEnd + 2)
|
||||
assert.match(lineRangeSource, /isManualPreviewItem\(item\)[\s\S]*?手动新增，无源文件定位/, '来源标签仍会把缺少行偏移的正常记录误判为手动新增')
|
||||
assert.match(lineRangeSource, /props\.processType === 'unstructured'[\s\S]*?来源：源文件记录/, '结构化来源记录缺少无行偏移时的准确标签')
|
||||
assert.doesNotMatch(lineRangeSource, /sourceStartLine == null[^\n]*手动新增/, '来源标签仍直接以缺少行号判定手动新增')
|
||||
assert.match(lineRangeSource, /sheet_name[\s\S]*?row_number[\s\S]*?来源：\$\{sheet\} · 第 \$\{locator\.row_number\} 行/, 'XLSX 来源标签没有展示工作表和物理行号')
|
||||
assert.match(lineRangeSource, /json_pointer[\s\S]*?JSON 路径/, 'JSON 来源标签没有展示 JSON 路径')
|
||||
assert.match(lineRangeSource, /locator\?\.kind === 'json'[\s\S]*?JSON 根对象/, 'JSON 根对象来源标签被空 JSON Pointer 错误降级')
|
||||
assert.match(lineRangeSource, /locatedLines[\s\S]*?第 \$\{locatedLines\.start\}[\s\S]*?locatedLines\.end/, 'JSONL/CSV 来源标签没有展示行范围')
|
||||
assert.match(lineRangeSource, /headingPath[\s\S]*?章节：/, '非结构化来源标签没有合并标题路径')
|
||||
assert.match(previewSource, /sourceLocator\?\.start_line[\s\S]*?sourceLocator\?\.end_line/, '文本预览没有优先使用后端行号定位')
|
||||
assert.match(previewSource, /sourceLocator\?\.source_start\s*\?\?\s*item\.sourceStart/, '文本预览没有优先使用 locator 字符起点')
|
||||
assert.match(previewSource, /sourceLocator\?\.source_end\s*\?\?\s*item\.sourceEnd/, '文本预览没有优先使用 locator 字符终点')
|
||||
assert.match(previewSource, /data-line-number="line\.number"/, '文本预览行缺少稳定行号定位标识')
|
||||
assert.match(previewSource, /isLineHighlighted\(line\.number, line\.start, line\.end\)/, '文本预览没有按物理行号高亮')
|
||||
assert.match(previewSource, /querySelector<HTMLElement>\(`\[data-line-number=/, '选中记录后没有按物理行号滚动定位')
|
||||
assert.match(previewSource, /const SOURCE_LINE_RENDER_LIMIT = 240/, '源文件查看器缺少安全渲染上限')
|
||||
assert.match(previewSource, /const SOURCE_LINE_CHARACTER_LIMIT = 4_000/, '源文件查看器缺少单行字符渲染上限')
|
||||
assert.match(previewSource, /sourceLineWindow\([\s\S]*?SOURCE_LINE_RENDER_LIMIT/, '源文件查看器没有使用有界行窗口')
|
||||
assert.match(previewSource, /SOURCE_LINE_RENDER_LIMIT,[\s\S]*?SOURCE_LINE_CHARACTER_LIMIT,[\s\S]*?selectedSourceLine\.value,[\s\S]*?selectedSourceOffset\.value/, '单行超大 JSON 没有围绕选中来源构建字符窗口')
|
||||
assert.match(previewSource, /sourceWindowStartLine/, '源文件查看器缺少窗口起始行状态')
|
||||
assert.match(previewSource, /showPreviousSourceWindow[\s\S]*?showNextSourceWindow/, '源文件查看器缺少前后窗口导航')
|
||||
assert.match(previewSource, /sourceLineNumberAtOffset\(props\.sourceText/, '仅有字符偏移时没有解析目标物理行')
|
||||
assert.match(previewSource, /filterable/, '文件选择器必须可搜索')
|
||||
assert.match(previewSource, /当前文件/, '预览缺少当前文件切换器')
|
||||
assert.doesNotMatch(previewSource, /located-badge|sync-label|已定位到/, '源文件栏不应显示冗余定位提示')
|
||||
@@ -269,6 +321,18 @@ for (const marker of [
|
||||
]) {
|
||||
assert.ok(officeViewerSource.includes(marker), `Word/XLSX 预览缺少结构或行为：${marker}`)
|
||||
}
|
||||
assert.match(officeViewerSource, /const selectedXlsxLocator = computed/, 'XLSX 查看器没有读取精确来源定位')
|
||||
assert.match(officeViewerSource, /row\.row_number === locator\.row_number/, 'XLSX 查看器没有按物理行号精确高亮')
|
||||
assert.match(officeViewerSource, /row\.record_index === locator\.sheet_record_index/, 'XLSX 查看器没有按工作表记录序号精确高亮')
|
||||
assert.match(officeViewerSource, /Math\.floor\(locator\.sheet_record_index \/ XLSX_PAGE_SIZE\) \* XLSX_PAGE_SIZE/, 'XLSX 查看器没有按记录序号自动计算分页')
|
||||
assert.match(officeViewerSource, /activeSheetIndex\.value = targetSheet[\s\S]*?pageOffset\.value = targetOffset[\s\S]*?loadPreview\(\)/, '切换记录时 XLSX 查看器没有自动切工作表和分页')
|
||||
const xlsxHighlightStart = officeViewerSource.indexOf('function xlsxRowHighlighted(')
|
||||
const xlsxHighlightEnd = officeViewerSource.indexOf('\n}', xlsxHighlightStart)
|
||||
const xlsxHighlightSource = officeViewerSource.slice(xlsxHighlightStart, xlsxHighlightEnd + 2)
|
||||
assert.ok(
|
||||
xlsxHighlightSource.indexOf('locator.row_number') < xlsxHighlightSource.indexOf('selectedRecordKey.value'),
|
||||
'XLSX 查看器没有把精确定位放在原内容比对 fallback 之前',
|
||||
)
|
||||
|
||||
const taskSetupPath = path.join(createDir, 'TaskSetupStep.vue')
|
||||
const structuredOptionsPath = path.join(createDir, 'StructuredOptionsPanel.vue')
|
||||
@@ -454,7 +518,7 @@ assert.match(
|
||||
)
|
||||
assert.match(
|
||||
workflowInitializationSource,
|
||||
/sourceTask\.status === 'running'[\s\S]*?resumeStep = 'generate'[\s\S]*?goToStep\(resumeStep\)[\s\S]*?resumeGeneration/,
|
||||
/sourceTask\.status === 'running'[\s\S]*?resumeStep = 'generate'[\s\S]*?resumeGeneration\(\)[\s\S]*?goToStep\(resumeStep\)/,
|
||||
'生成运行中时没有强制回到第五步并接管后台进度',
|
||||
)
|
||||
const startGenerationHandler = viewSource.slice(
|
||||
@@ -463,7 +527,35 @@ const startGenerationHandler = viewSource.slice(
|
||||
)
|
||||
assert.match(startGenerationHandler, /await persistWorkflowStep\('generate'\)[\s\S]*?await startGeneration\(\)[\s\S]*?dirty\.value = false/, '开始生成没有持久化第五步或启动真实后台任务')
|
||||
assert.doesNotMatch(startGenerationHandler, /router\.(?:push|replace)|allowLeave\s*=\s*true/, '开始生成后应停留在第五步，不得自动跳回列表')
|
||||
assert.match(viewSource, /:disabled="currentStepId === 'generate' \|\| previewBuilding \|\| sourceUploading"/, '第五步底部返回按钮没有固定禁用')
|
||||
assert.match(
|
||||
generationSource,
|
||||
/const canReturnFromGeneration = computed\(\(\) => \([\s\S]*?generation\.status === 'idle'[\s\S]*?!generationStarting\.value[\s\S]*?!generationRestoring\.value/,
|
||||
'第五步返回权限没有区分未启动、启动中和恢复中状态',
|
||||
)
|
||||
assert.match(
|
||||
viewSource,
|
||||
/:disabled="\(currentStepId === 'generate' && !canReturnFromGeneration\) \|\| previewBuilding \|\| sourceUploading"/,
|
||||
'第五步尚未启动生成时返回按钮仍被禁用',
|
||||
)
|
||||
const handleBackStart = viewSource.indexOf('async function handleBack()')
|
||||
const handleBackEnd = viewSource.indexOf('\n}', handleBackStart)
|
||||
const handleBackSource = viewSource.slice(handleBackStart, handleBackEnd + 2)
|
||||
assert.match(
|
||||
handleBackSource,
|
||||
/currentStepId\.value === 'generate' && !canReturnFromGeneration\.value/,
|
||||
'第五步处理函数仍无条件拦截返回',
|
||||
)
|
||||
assert.match(
|
||||
generationSource,
|
||||
/async function resumeGeneration\(\)[\s\S]*?generationRestoring\.value = true[\s\S]*?await getDataProcessProgress\(taskId\)[\s\S]*?generationRestoring\.value = false/,
|
||||
'恢复已启动任务时存在短暂可返回的 idle 窗口',
|
||||
)
|
||||
assert.match(viewSource, /const resume = resumeGeneration\(\)[\s\S]*?goToStep\(resumeStep\)[\s\S]*?await resume/, '第五步展示时未先启动恢复锁')
|
||||
assert.match(
|
||||
generationSource,
|
||||
/const generationStarting = ref\(false\)[\s\S]*?generationStarting\.value = true[\s\S]*?generationStarting\.value = false/,
|
||||
'点击开始生成后到请求启动前没有锁定返回状态',
|
||||
)
|
||||
assert.match(
|
||||
viewSource,
|
||||
/generation\.status === 'success'[\s\S]*?persistWorkflowStep\('results'\)/,
|
||||
@@ -506,32 +598,34 @@ assert.match(viewSource, /watch\(processType,[\s\S]*?resetSourceDataForProcessTy
|
||||
assert.match(viewSource, /function resetSourceDataForProcessTypeChange\(\)[\s\S]*?uploadedFiles\.value = \[\][\s\S]*?selectedPreviewFileId\.value = null/, '旧源数据失效没有同步清理文件与预览选择')
|
||||
|
||||
assert.match(taskSetupSource, /v-if="processType === 'structured'"/, '结构化配置必须仅在结构化数据类型下显示')
|
||||
const expectedStructuredOptions = [
|
||||
['clean_invalid', '清理无效数据', '清理全空列，并剔除关键字段残缺的数据行'],
|
||||
const expectedStructuredGroups = [
|
||||
[
|
||||
'detect_structure',
|
||||
'嵌套结构展平',
|
||||
'展平嵌套对象和可解析的 JSON 字段；Excel 表头与合并单元格在上传时自动解析',
|
||||
"values: ['clean_invalid', 'deduplicate']",
|
||||
'数据清洗',
|
||||
'清理全空列和空记录，并删除内容完全相同的记录；不会猜测可空字段是否必填',
|
||||
],
|
||||
[
|
||||
'deduplicate',
|
||||
'重复记录去重',
|
||||
'按整行内容或 id、uuid、key、code、*_id 等身份字段去重，暂不支持自定义组合字段',
|
||||
"values: ['detect_structure', 'normalize_format']",
|
||||
'结构标准化',
|
||||
'展平嵌套对象和可解析的 JSON 字段，并统一编码、空白、字段名和 JSON 序列化格式',
|
||||
],
|
||||
['normalize_format', '数据格式标准化', '按所选规则统一编码、空白、字段名及 JSON 序列化格式'],
|
||||
['filter_anomaly', '异常数据过滤', '使用 IQR 识别数值离群值，并过滤乱码等异常记录'],
|
||||
['desensitize', '敏感信息脱敏', '识别并脱敏姓名、手机号、邮箱和身份证号'],
|
||||
["values: ['desensitize']", '敏感信息脱敏', '识别并脱敏姓名、手机号、邮箱和身份证号'],
|
||||
]
|
||||
for (const [value, label, description] of expectedStructuredOptions) {
|
||||
assert.ok(structuredOptionsSource.includes(`value: '${value}'`), `结构化预处理缺少值：${value}`)
|
||||
for (const [values, label, description] of expectedStructuredGroups) {
|
||||
assert.ok(structuredOptionsSource.includes(values), `结构化预处理组合值不准确：${label}`)
|
||||
assert.ok(structuredOptionsSource.includes(`label: '${label}'`), `结构化预处理缺少标签：${label}`)
|
||||
assert.ok(structuredOptionsSource.includes(`description: '${description}'`), `结构化预处理语义不准确：${value}`)
|
||||
assert.ok(structuredOptionsSource.includes(`description: '${description}'`), `结构化预处理语义不准确：${label}`)
|
||||
}
|
||||
const structuredOptionValues = [...structuredOptionsSource.matchAll(/\{\s*value: '([^']+)',\s*label:/g)]
|
||||
.map((match) => match[1])
|
||||
assert.deepEqual(structuredOptionValues, expectedStructuredOptions.map(([value]) => value), '结构化预处理值集合不准确')
|
||||
assert.equal(new Set(structuredOptionValues).size, structuredOptionValues.length, '结构化预处理 value 必须唯一')
|
||||
assert.match(structuredOptionsSource, /Array\.from\(new Set\(value\.filter\(/, '结构化预处理选中值没有去重')
|
||||
assert.equal(expectedStructuredGroups.length, 3, '结构化预处理应收敛为 3 项')
|
||||
const preprocessGroupsSource = structuredOptionsSource.slice(
|
||||
structuredOptionsSource.indexOf('const PREPROCESS_GROUPS'),
|
||||
structuredOptionsSource.indexOf('const legacyAnomalyFilterEnabled'),
|
||||
)
|
||||
assert.doesNotMatch(preprocessGroupsSource, /异常数据过滤|filter_anomaly|IQR/, '结构化新任务仍暴露异常数据过滤')
|
||||
assert.match(structuredOptionsSource, /:indeterminate="groupIndeterminate\(group\.values\)"/, '历史部分选中的组合项没有半选回显')
|
||||
assert.match(structuredOptionsSource, /function updatePreprocessGroup\([\s\S]*?new Set\(props\.options\.preprocessOptions\)[\s\S]*?next\.add\(value\)[\s\S]*?next\.delete\(value\)[\s\S]*?\[\.\.\.next\]/, '结构化预处理组合开关没有原子化更新或去重内部选项')
|
||||
assert.match(typesSource, /仅用于恢复历史任务[\s\S]*?\| 'filter_anomaly'/, '异常数据过滤缺少历史兼容类型')
|
||||
assert.match(structuredOptionsSource, /legacyAnomalyFilterEnabled[\s\S]*?历史任务[\s\S]*?结果可复现/, '历史异常过滤配置没有透明提示')
|
||||
assert.ok(structuredOptionsSource.includes('生成选项'), '结构化配置缺少生成选项分类')
|
||||
for (const splitName of ['训练集', '验证集', '测试集']) {
|
||||
assert.ok(datasetSplitEditorSource.includes(splitName), `生成选项缺少数据集划分：${splitName}`)
|
||||
@@ -585,8 +679,20 @@ for (const extension of ['txt', 'md', 'markdown', 'pdf', 'docx', 'pptx', 'json',
|
||||
}
|
||||
assert.match(sourceUploadWorkerSource, /LEGACY_OFFICE_EXTENSIONS = new Set\(\['doc', 'xls', 'ppt'\]\)/, '缺少旧版 Office 格式识别')
|
||||
assert.ok(sourceUploadWorkerSource.includes('请分别转换为 DOCX、XLSX、PPTX 后上传'), '旧版 Office 文件缺少转换提示')
|
||||
assert.match(sourceUploadWorkerSource, /if \(!BINARY_FILE_EXTENSIONS\.has\(job\.extension\)\) \{[\s\S]*?TextDecoder/, '文本格式没有执行 UTF-8 客户端校验')
|
||||
assert.match(sourceUploadWorkerSource, /if \(BINARY_FILE_EXTENSIONS\.has\(job\.extension\)\) \{[\s\S]*?getDataProcessSourceContent\(currentTaskId, source\.id,[\s\S]*?start_line:\s*1,[\s\S]*?line_count:\s*10_000/, '二进制文档上传后没有读取后端解析文本')
|
||||
const sourceValidationStart = sourceUploadWorkerSource.indexOf('export function validateSourceFileSelection(')
|
||||
const sourceValidationEnd = sourceUploadWorkerSource.indexOf('\n}\n\nfunction unicodeCodePointLength', sourceValidationStart)
|
||||
assert.ok(sourceValidationStart >= 0 && sourceValidationEnd > sourceValidationStart, '无法定位源文件选择校验函数')
|
||||
const sourceValidationSource = sourceUploadWorkerSource.slice(sourceValidationStart, sourceValidationEnd + 2)
|
||||
assert.doesNotMatch(sourceValidationSource, /file\.name === raw\.name[\s\S]{0,160}file\.size === raw\.size|同名且同大小/, '不同内容但同名同大小的文件仍会被前端误拒绝')
|
||||
assert.match(sourceValidationSource, /selectedFiles\.length >= MAX_SOURCE_FILE_COUNT/, '移除伪重复校验时误删了文件数量限制')
|
||||
assert.match(sourceValidationSource, /selectedBytes \+ raw\.size > MAX_SOURCE_BATCH_BYTES/, '移除伪重复校验时误删了批次大小限制')
|
||||
assert.doesNotMatch(sourceUploadWorkerSource, /job\.file\.arrayBuffer\(|new TextDecoder/, '上传前仍把整个文本文件读入浏览器内存')
|
||||
assert.match(sourceUploadWorkerSource, /export async function loadCanonicalSourceContent[\s\S]*?offset,[\s\S]*?limit: SOURCE_CONTENT_PAGE_CHARS/, '服务端 canonical content 没有按有界字符窗口读取')
|
||||
assert.match(sourceUploadWorkerSource, /pending\.content = await loadCanonicalSourceContent\(currentTaskId, source\.id\)/, '上传成功后没有统一使用服务端 canonical content')
|
||||
assert.doesNotMatch(sourceUploadWorkerSource, /\brawFile:\s*job\.file\b/, '上传成功状态仍长期保留原始 File')
|
||||
assert.doesNotMatch(typesSource, /\brawFile\??:\s*File\b/, '上传状态类型仍长期持有原始 File')
|
||||
assert.doesNotMatch(viewSource, /\brawFile:\s*raw\b/, '待上传列表仍复制保存原始 File')
|
||||
assert.match(apiSource, /params:\s*\{[\s\S]*?offset\?: number[\s\S]*?limit\?: number[\s\S]*?\}/, '正文 API 前端契约缺少字符窗口参数')
|
||||
assert.match(apiSource, /formData\.append\('files', file\)/, '上传 API 没有使用 files 多文件表单字段')
|
||||
assert.match(apiSource, /onUploadProgress:[\s\S]*?event\.loaded \/ event\.total[\s\S]*?Math\.min\(99,/, '上传 API 没有接入真实字节进度或响应前未限制在 99%')
|
||||
assert.match(apiSource, /source-files`[\s\S]*?timeout: 5 \* 60 \* 1000/, '源文件上传缺少 5 分钟超时')
|
||||
@@ -609,7 +715,7 @@ assert.match(
|
||||
/export interface DataProcessPreviewProgress[\s\S]*?workflow_step: DataProcessWorkflowStep[\s\S]*?preview_status: DataProcessPreviewStatus[\s\S]*?preview_progress: number[\s\S]*?preview_run_id/,
|
||||
'后台切分进度契约缺少步骤、状态、进度或任务代次',
|
||||
)
|
||||
for (const field of ['rawFile', 'status', 'uploadProgress', 'uploadError', 'previewStatus', 'previewProgress', 'previewError', 'previewConfigSignature']) {
|
||||
for (const field of ['status', 'uploadProgress', 'uploadError', 'previewStatus', 'previewProgress', 'previewError', 'previewConfigSignature']) {
|
||||
assert.ok(typesSource.includes(field), `上传文件缺少逐文件预览字段：${field}`)
|
||||
}
|
||||
assert.match(typesSource, /status: 'queued' \| 'uploading' \| 'ready' \| 'failed'/, '上传文件状态机不完整')
|
||||
@@ -853,13 +959,23 @@ const defaultStructuredPreprocess = defaultPreprocessValues(
|
||||
)
|
||||
assert.deepEqual(
|
||||
defaultStructuredPreprocess,
|
||||
['clean_invalid', 'detect_structure', 'deduplicate', 'normalize_format'],
|
||||
'结构化默认预处理配置不准确',
|
||||
[],
|
||||
'结构化新任务不应默认勾选预处理',
|
||||
)
|
||||
assert.equal(new Set(defaultStructuredPreprocess).size, defaultStructuredPreprocess.length, '结构化默认预处理值重复')
|
||||
const defaultUnstructuredPreprocess = defaultPreprocessValues('createDefaultUnstructuredOptions')
|
||||
assert.deepEqual(defaultUnstructuredPreprocess, expectedSmartPreprocessOptions, '智能预处理默认值不完整')
|
||||
assert.deepEqual(defaultUnstructuredPreprocess, [], '非结构化新任务不应默认勾选预处理')
|
||||
assert.equal(new Set(defaultUnstructuredPreprocess).size, defaultUnstructuredPreprocess.length, '非结构化默认预处理值重复')
|
||||
for (const field of ['preserveTables', 'preserveCodeBlocks', 'preserveLists']) {
|
||||
assert.match(
|
||||
stateSource,
|
||||
new RegExp(`${field}:\\s*false`),
|
||||
`非结构化预处理选项 ${field} 不应默认开启`,
|
||||
)
|
||||
}
|
||||
assert.match(structuredOptionsSource, /默认不执行预处理，请按数据情况自行选择/, '结构化预处理缺少默认不勾选说明')
|
||||
assert.match(unstructuredOptionsSource, /默认不执行预处理，请按文档情况自行选择/, '非结构化预处理缺少默认不勾选说明')
|
||||
assert.doesNotMatch(unstructuredOptionsSource, /默认启用结构感知/, '非结构化预处理仍保留默认启用的误导文案')
|
||||
|
||||
const backendConfigStart = viewSource.indexOf('function toBackendConfig()')
|
||||
const backendConfigEnd = viewSource.indexOf('function taskPayload()', backendConfigStart)
|
||||
@@ -934,7 +1050,7 @@ for (const [field, fallback] of [
|
||||
)
|
||||
}
|
||||
assert.match(regenerationSource, /getDataProcessTask\(sourceTaskId\.value\)/, '重新生成没有加载原任务')
|
||||
assert.match(regenerationSource, /while \(true\)[\s\S]*?getDataProcessSourceContent[\s\S]*?has_more/, '重新生成没有分页加载完整源正文')
|
||||
assert.match(regenerationSource, /loadCanonicalSourceContent\(taskId, file\.id\)/, '重新生成没有复用分页 canonical 正文加载器')
|
||||
assert.match(regenerationSource, /getDataProcessPreview\(taskId, \{ page: 1, page_size: 500 \}\)[\s\S]*?for \(let page = 2; page <= pages;/, '重新生成没有分页加载全部现有切片')
|
||||
assert.match(viewSource, /if \(hydrating\.value\) return/, '任务水合期间仍可能触发重置副作用')
|
||||
assert.match(regenerationSource, /currentSignature !== originalPreviewConfigSignature\.value[\s\S]*?currentSignature === confirmedPreviewConfigSignature\.value/, '切分变更确认没有按原签名和已确认签名去重')
|
||||
@@ -948,7 +1064,7 @@ assert.match(regenerationSource, /if \(regenerationPrepared\.value\) \{[\s\S]*?g
|
||||
assert.match(regenerationSource, /regenerationPrepared\.value = true/, '重新生成提交成功后没有记录服务端已变更状态')
|
||||
assert.match(regenerationSource, /hydrateWorkspace\(regeneratedTask, !regenerated\.preview_invalidated\)/, '重新生成没有按 preview_invalidated 决定保留或清空切片')
|
||||
assert.match(regenerationSource, /重新生成配置已保存，但工作区恢复失败/, '重新生成配置已保存但水合失败时缺少可恢复错误状态')
|
||||
assert.match(regenerationSource, /return chunks\.join\(''\)/, '分页恢复源正文时不应额外插入换行')
|
||||
assert.match(sourceUploadWorkerSource, /return chunks\.join\(''\)/, '分页恢复源正文时不应额外插入换行')
|
||||
assert.doesNotMatch(regenerationSource, /binaryDocument[\s\S]*?mapDataProcessSourceFile\(file, ''\)/, '二进制源正文加载失败时不能静默降级为空内容')
|
||||
assert.match(nextFromModelSource, /if \(isRegeneration\.value\) \{[\s\S]*?prepareRegeneration\(taskPayload\(\)\)/, '重新生成每次从模型步骤继续时没有调用专用接口')
|
||||
assert.doesNotMatch(nextFromModelSource, /isRegeneration\.value && !taskId\.value/, '重新生成提交一次后可能错误转为普通任务更新')
|
||||
@@ -1017,6 +1133,17 @@ for (const mutationFunction of [
|
||||
const mutationSource = viewSource.slice(mutationStart, mutationEnd === -1 ? undefined : mutationEnd)
|
||||
assert.ok(mutationSource.includes('resetDownstream()'), `预览变更 ${mutationFunction} 后没有失效旧生成结果`)
|
||||
}
|
||||
const updatePreviewContentStart = viewSource.indexOf('function updatePreviewContent(')
|
||||
const updatePreviewContentEnd = viewSource.indexOf('\n}', updatePreviewContentStart)
|
||||
const updatePreviewContentSource = viewSource.slice(updatePreviewContentStart, updatePreviewContentEnd + 2)
|
||||
assert.match(updatePreviewContentSource, /isManualPreviewItem\(item\)/, '编辑预览内容仍未按稳定来源信息区分手动项')
|
||||
assert.doesNotMatch(updatePreviewContentSource, /sourceStart == null/, '结构化来源记录编辑后仍会被误标为手动项')
|
||||
const restorePreviewItemStart = viewSource.indexOf('function restorePreviewItem(')
|
||||
const restorePreviewItemEnd = viewSource.indexOf('\n}', restorePreviewItemStart)
|
||||
const restorePreviewItemSource = viewSource.slice(restorePreviewItemStart, restorePreviewItemEnd + 2)
|
||||
assert.match(restorePreviewItemSource, /isManualPreviewItem\(item\)/, '恢复预览内容没有使用统一的手动项判定')
|
||||
assert.doesNotMatch(restorePreviewItemSource, /sourceStart == null/, '结构化来源记录仍因缺少字符偏移而无法恢复')
|
||||
assert.match(previewSource, /v-if="!isManualPreviewItem\(editingItem\)"/, '结构化来源记录的恢复原文按钮仍被错误隐藏')
|
||||
assert.doesNotMatch(modelSource, /createResults\(/, '纯预览映射模块不应承担结果生成职责')
|
||||
|
||||
function findNextStyleBlockStart(source, startIndex) {
|
||||
|
||||
Reference in New Issue
Block a user