md-spreadsheet-parser 1.2.2 → 1.4.0

package/package.json

@@ -1,6 +1,6 @@
 {
     "name": "md-spreadsheet-parser",
-    "version": "1.2.2",
+    "version": "1.4.0",
     "description": "A robust Markdown table parser and manipulator, powered by Python and WebAssembly.",
     "main": "dist/index.js",
     "types": "dist/index.d.ts",
@@ -24,7 +24,7 @@
     "author": "f-y",
     "repository": {
         "type": "git",
-        "url": "git+https://github.com/f-y/md-spreadsheet-parser.git"
+        "url": "git+https://github.com/fy-labs/md-spreadsheet-parser.git"
     },
     "license": "MIT",
     "devDependencies": {

package/src/generated_adapter.py

@@ -67,6 +67,8 @@ def unwrap_table(obj: Any) -> Any:
 class WitSheet:
     name: Any = None
     tables: Any = None
+    sheet_type: Any = None
+    content: Any = None
     metadata: Any = None
 
 def convert_sheet(obj: Any) -> WitSheet:
@@ -74,6 +76,8 @@ def convert_sheet(obj: Any) -> WitSheet:
     res = WitSheet()
     res.name = obj.name
     res.tables = [convert_table(x) for x in obj.tables]
+    res.sheet_type = str(obj.sheet_type) if obj.sheet_type is not None else None
+    res.content = obj.content
     res.metadata = json.dumps(obj.metadata or {}) if obj.metadata is not None else None
     return res
 
@@ -84,6 +88,10 @@ def unwrap_sheet(obj: Any) -> Any:
         kwargs['name'] = obj.name
     if obj.tables is not None:
         kwargs['tables'] = [unwrap_table(x) for x in obj.tables]
+    if obj.sheet_type is not None:
+        kwargs['sheet_type'] = obj.sheet_type
+    if obj.content is not None:
+        kwargs['content'] = obj.content
     if obj.metadata is not None:
         kwargs['metadata'] = json.loads(obj.metadata)
     return models.Sheet(**kwargs)
@@ -91,13 +99,21 @@ def unwrap_sheet(obj: Any) -> Any:
 @dataclass
 class WitWorkbook:
     sheets: Any = None
+    name: Any = None
+    start_line: Any = None
+    end_line: Any = None
     metadata: Any = None
+    root_content: Any = None
 
 def convert_workbook(obj: Any) -> WitWorkbook:
     if obj is None: return None
     res = WitWorkbook()
     res.sheets = [convert_sheet(x) for x in obj.sheets]
+    res.name = obj.name
+    res.start_line = obj.start_line
+    res.end_line = obj.end_line
     res.metadata = json.dumps(obj.metadata or {}) if obj.metadata is not None else None
+    res.root_content = obj.root_content
     return res
 
 def unwrap_workbook(obj: Any) -> Any:
@@ -105,8 +121,16 @@ def unwrap_workbook(obj: Any) -> Any:
     kwargs = {}
     if obj.sheets is not None:
         kwargs['sheets'] = [unwrap_sheet(x) for x in obj.sheets]
+    if obj.name is not None:
+        kwargs['name'] = obj.name
+    if obj.start_line is not None:
+        kwargs['start_line'] = obj.start_line
+    if obj.end_line is not None:
+        kwargs['end_line'] = obj.end_line
     if obj.metadata is not None:
         kwargs['metadata'] = json.loads(obj.metadata)
+    if obj.root_content is not None:
+        kwargs['root_content'] = obj.root_content
     return models.Workbook(**kwargs)
 
 @dataclass

package/src/index.ts

@@ -144,16 +144,19 @@ export class Table {
     startLine: number | undefined;
     endLine: number | undefined;
 
-    constructor(data?: Partial<Table>) {
+    constructor(data?: Partial<Table> & Record<string, any>) {
         if (data) {
-            this.headers = data.headers;
-            this.rows = data.rows;
-            this.alignments = data.alignments;
-            this.name = data.name;
-            this.description = data.description;
-            this.metadata = (typeof data.metadata === 'string') ? JSON.parse(data.metadata) : data.metadata;
-            this.startLine = data.startLine;
-            this.endLine = data.endLine;
+            this.headers = data.headers ?? data.headers;
+            this.rows = data.rows ?? data.rows;
+            this.alignments = data.alignments ?? data.alignments;
+            this.name = data.name ?? data.name;
+            this.description = data.description ?? data.description;
+            {
+                const val = data.metadata ?? data.metadata;
+                this.metadata = (typeof val === 'string') ? JSON.parse(val) : val;
+            }
+            this.startLine = data.startLine ?? data.start_line;
+            this.endLine = data.endLine ?? data.end_line;
         }
     }
 
@@ -274,13 +277,20 @@ export class Table {
 export class Sheet {
     name: string | undefined;
     tables: any[] | undefined;
+    sheetType: string | undefined;
+    content: any | undefined;
     metadata: any | undefined;
 
-    constructor(data?: Partial<Sheet>) {
+    constructor(data?: Partial<Sheet> & Record<string, any>) {
         if (data) {
-            this.name = data.name;
-            this.tables = (data.tables || []).map((x: any) => x instanceof Table ? x : new Table(x));
-            this.metadata = (typeof data.metadata === 'string') ? JSON.parse(data.metadata) : data.metadata;
+            this.name = data.name ?? data.name;
+            this.tables = ((data.tables ?? data.tables) || []).map((x: any) => x instanceof Table ? x : new Table(x));
+            this.sheetType = data.sheetType ?? data.sheet_type;
+            this.content = data.content ?? data.content;
+            {
+                const val = data.metadata ?? data.metadata;
+                this.metadata = (typeof val === 'string') ? JSON.parse(val) : val;
+            }
         }
     }
 
@@ -299,6 +309,8 @@ export class Sheet {
         return {
             name: this.name,
             tables: (this.tables || []).map((t: any) => t.json ? t.json : t),
+            sheetType: this.sheetType,
+            content: this.content,
             metadata: this.metadata ?? {},
         };
     }
@@ -360,12 +372,23 @@ export class Sheet {
 
 export class Workbook {
     sheets: any[] | undefined;
+    name: string | undefined;
+    startLine: number | undefined;
+    endLine: number | undefined;
     metadata: any | undefined;
+    rootContent: any | undefined;
 
-    constructor(data?: Partial<Workbook>) {
+    constructor(data?: Partial<Workbook> & Record<string, any>) {
         if (data) {
-            this.sheets = (data.sheets || []).map((x: any) => x instanceof Sheet ? x : new Sheet(x));
-            this.metadata = (typeof data.metadata === 'string') ? JSON.parse(data.metadata) : data.metadata;
+            this.sheets = ((data.sheets ?? data.sheets) || []).map((x: any) => x instanceof Sheet ? x : new Sheet(x));
+            this.name = data.name ?? data.name;
+            this.startLine = data.startLine ?? data.start_line;
+            this.endLine = data.endLine ?? data.end_line;
+            {
+                const val = data.metadata ?? data.metadata;
+                this.metadata = (typeof val === 'string') ? JSON.parse(val) : val;
+            }
+            this.rootContent = data.rootContent ?? data.root_content;
         }
     }
 
@@ -382,8 +405,10 @@ export class Workbook {
      */
     get json(): any {
         return {
+            name: this.name,
             sheets: (this.sheets || []).map((s: any) => s.json ? s.json : s),
             metadata: this.metadata ?? {},
+            rootContent: this.rootContent,
         };
     }
 
@@ -449,13 +474,13 @@ export class ParsingSchema {
     stripWhitespace: boolean | undefined;
     convertBrToNewline: boolean | undefined;
 
-    constructor(data?: Partial<ParsingSchema>) {
+    constructor(data?: Partial<ParsingSchema> & Record<string, any>) {
         if (data) {
-            this.columnSeparator = data.columnSeparator;
-            this.headerSeparatorChar = data.headerSeparatorChar;
-            this.requireOuterPipes = data.requireOuterPipes;
-            this.stripWhitespace = data.stripWhitespace;
-            this.convertBrToNewline = data.convertBrToNewline;
+            this.columnSeparator = data.columnSeparator ?? data.column_separator;
+            this.headerSeparatorChar = data.headerSeparatorChar ?? data.header_separator_char;
+            this.requireOuterPipes = data.requireOuterPipes ?? data.require_outer_pipes;
+            this.stripWhitespace = data.stripWhitespace ?? data.strip_whitespace;
+            this.convertBrToNewline = data.convertBrToNewline ?? data.convert_br_to_newline;
         }
     }
 
@@ -471,22 +496,22 @@ export class MultiTableParsingSchema {
     requireOuterPipes: boolean | undefined;
     stripWhitespace: boolean | undefined;
     convertBrToNewline: boolean | undefined;
-    rootMarker: string | undefined;
+    rootMarker: any | undefined;
     sheetHeaderLevel: number | undefined;
     tableHeaderLevel: number | undefined;
     captureDescription: boolean | undefined;
 
-    constructor(data?: Partial<MultiTableParsingSchema>) {
+    constructor(data?: Partial<MultiTableParsingSchema> & Record<string, any>) {
         if (data) {
-            this.columnSeparator = data.columnSeparator;
-            this.headerSeparatorChar = data.headerSeparatorChar;
-            this.requireOuterPipes = data.requireOuterPipes;
-            this.stripWhitespace = data.stripWhitespace;
-            this.convertBrToNewline = data.convertBrToNewline;
-            this.rootMarker = data.rootMarker;
-            this.sheetHeaderLevel = data.sheetHeaderLevel;
-            this.tableHeaderLevel = data.tableHeaderLevel;
-            this.captureDescription = data.captureDescription;
+            this.columnSeparator = data.columnSeparator ?? data.column_separator;
+            this.headerSeparatorChar = data.headerSeparatorChar ?? data.header_separator_char;
+            this.requireOuterPipes = data.requireOuterPipes ?? data.require_outer_pipes;
+            this.stripWhitespace = data.stripWhitespace ?? data.strip_whitespace;
+            this.convertBrToNewline = data.convertBrToNewline ?? data.convert_br_to_newline;
+            this.rootMarker = data.rootMarker ?? data.root_marker;
+            this.sheetHeaderLevel = data.sheetHeaderLevel ?? data.sheet_header_level;
+            this.tableHeaderLevel = data.tableHeaderLevel ?? data.table_header_level;
+            this.captureDescription = data.captureDescription ?? data.capture_description;
         }
     }
 
@@ -501,11 +526,17 @@ export class ConversionSchema {
     customConverters: string | undefined;
     fieldConverters: string | undefined;
 
-    constructor(data?: Partial<ConversionSchema>) {
+    constructor(data?: Partial<ConversionSchema> & Record<string, any>) {
         if (data) {
-            this.booleanPairs = data.booleanPairs;
-            this.customConverters = (typeof data.customConverters === 'string') ? JSON.parse(data.customConverters) : data.customConverters;
-            this.fieldConverters = (typeof data.fieldConverters === 'string') ? JSON.parse(data.fieldConverters) : data.fieldConverters;
+            this.booleanPairs = data.booleanPairs ?? data.boolean_pairs;
+            {
+                const val = data.customConverters ?? data.custom_converters;
+                this.customConverters = (typeof val === 'string') ? JSON.parse(val) : val;
+            }
+            {
+                const val = data.fieldConverters ?? data.field_converters;
+                this.fieldConverters = (typeof val === 'string') ? JSON.parse(val) : val;
+            }
         }
     }
 
@@ -523,12 +554,12 @@ export class ExcelParsingSchema {
     delimiter: string | undefined;
     headerSeparator: string | undefined;
 
-    constructor(data?: Partial<ExcelParsingSchema>) {
+    constructor(data?: Partial<ExcelParsingSchema> & Record<string, any>) {
         if (data) {
-            this.headerRows = data.headerRows;
-            this.fillMergedHeaders = data.fillMergedHeaders;
-            this.delimiter = data.delimiter;
-            this.headerSeparator = data.headerSeparator;
+            this.headerRows = data.headerRows ?? data.header_rows;
+            this.fillMergedHeaders = data.fillMergedHeaders ?? data.fill_merged_headers;
+            this.delimiter = data.delimiter ?? data.delimiter;
+            this.headerSeparator = data.headerSeparator ?? data.header_separator;
         }
     }