informatica-python 1.0.0__py3-none-any.whl

informatica_python/utils/expression_converter.py

@@ -0,0 +1,128 @@
+import re
+
+
+INFA_FUNC_MAP = {
+    "IIF": "iif_expr",
+    "DECODE": "decode_expr",
+    "LTRIM": "ltrim",
+    "RTRIM": "rtrim",
+    "UPPER": "upper",
+    "LOWER": "lower",
+    "SUBSTR": "substr",
+    "LPAD": "lpad",
+    "RPAD": "rpad",
+    "TO_CHAR": "to_char",
+    "TO_DATE": "to_date",
+    "TO_INTEGER": "to_integer",
+    "TO_BIGINT": "to_bigint",
+    "TO_FLOAT": "to_float",
+    "TO_DECIMAL": "to_decimal",
+    "SYSDATE": "current_timestamp",
+    "SYSTIMESTAMP": "current_timestamp",
+    "SESSSTARTTIME": "session_start_time",
+    "GET_DATE_PART": "get_date_part",
+    "ADD_TO_DATE": "add_to_date",
+    "TRUNC": "trunc",
+    "ROUND": "round",
+    "ABS": "abs",
+    "CEIL": "ceil",
+    "FLOOR": "floor",
+    "MOD": "mod",
+    "POWER": "power",
+    "SQRT": "sqrt",
+    "LENGTH": "length",
+    "CONCAT": "concat",
+    "INSTR": "instr",
+    "REPLACECHR": "replacechr",
+    "REPLACESTR": "replacestr",
+    "REG_EXTRACT": "reg_extract",
+    "REG_MATCH": "reg_match",
+    "REG_REPLACE": "reg_replace",
+    "IS_DATE": "is_date",
+    "IS_NUMBER": "is_number",
+    "IS_SPACES": "is_spaces",
+    "NVL": "nvl",
+    "NVL2": "nvl2",
+    "ISNULL": "isnull",
+    "ERROR": "raise_error",
+    "ABORT": "abort_func",
+    "LOOKUP": "lookup_func",
+    "MAX": "max_val",
+    "MIN": "min_val",
+    "SUM": "sum_val",
+    "COUNT": "count_val",
+    "AVG": "avg_val",
+    "FIRST": "first_val",
+    "LAST": "last_val",
+    "MOVINGAVG": "moving_avg",
+    "MOVINGSUM": "moving_sum",
+    "CUME": "cume",
+    "SETCOUNTVARIABLE": "set_count_variable",
+    "SETVARIABLE": "set_variable",
+}
+
+
+def convert_expression(expr):
+    if not expr or not expr.strip():
+        return "None"
+
+    cleaned = expr.strip()
+
+    if re.match(r'^[A-Za-z_][A-Za-z0-9_]*$', cleaned):
+        return f'row["{cleaned}"]'
+
+    if re.match(r'^-?\d+(\.\d+)?$', cleaned):
+        return cleaned
+
+    if cleaned.startswith("'") and cleaned.endswith("'"):
+        return cleaned
+
+    converted = cleaned
+
+    for infa_func, py_func in INFA_FUNC_MAP.items():
+        pattern = re.compile(r'\b' + re.escape(infa_func) + r'\s*\(', re.IGNORECASE)
+        converted = pattern.sub(f'{py_func}(', converted)
+
+    converted = converted.replace("||", " + ")
+
+    converted = re.sub(r':LKP\.(\w+)\(', r'lookup_func("\1", ', converted)
+
+    converted = re.sub(r'\$\$(\w+)', r'get_variable("\1")', converted)
+
+    converted = re.sub(r':(\w+)', r'row["\1"]', converted)
+
+    return converted
+
+
+def convert_sql_expression(sql_expr):
+    if not sql_expr or not sql_expr.strip():
+        return ""
+
+    converted = sql_expr.strip()
+    converted = converted.replace("
", "\n")
+    converted = converted.replace("
", "\r")
+    converted = converted.replace("
", "\n")
+    converted = converted.replace("	", "\t")
+    converted = converted.replace("'", "'")
+    converted = converted.replace(""", '"')
+    converted = converted.replace("&", "&")
+    converted = converted.replace("&lt;", "<")
+    converted = converted.replace("&gt;", ">")
+
+    return converted
+
+
+def detect_sql_dialect(sql_text):
+    if not sql_text:
+        return "generic"
+
+    sql_upper = sql_text.upper()
+
+    if "GETDATE()" in sql_upper or "ISNULL(" in sql_upper or "TOP " in sql_upper:
+        return "mssql"
+    if "NVL(" in sql_upper or "SYSDATE" in sql_upper or "ROWNUM" in sql_upper:
+        return "oracle"
+    if "NOW()" in sql_upper or "COALESCE(" in sql_upper:
+        return "postgresql"
+
+    return "generic"

informatica_python-1.0.0.dist-info/METADATA

@@ -0,0 +1,118 @@
+Metadata-Version: 2.4
+Name: informatica-python
+Version: 1.0.0
+Summary: Convert Informatica PowerCenter workflow XML to Python/PySpark code
+License: MIT
+Requires-Python: >=3.8
+Description-Content-Type: text/markdown
+Requires-Dist: lxml>=4.9.0
+Requires-Dist: pyyaml>=6.0
+Provides-Extra: dev
+Requires-Dist: pytest>=7.0; extra == "dev"
+
+# informatica-python
+
+Convert Informatica PowerCenter workflow XML files to Python/PySpark code.
+
+## Installation
+
+```bash
+pip install informatica-python
+```
+
+## Quick Start
+
+### Command Line
+
+```bash
+# Convert XML to Python files in a directory
+informatica-python workflow.xml -o output_dir
+
+# Convert XML to a zip file
+informatica-python workflow.xml -z output.zip
+
+# Use a different data library (pandas, dask, polars, vaex, modin)
+informatica-python workflow.xml -o output_dir --data-lib polars
+
+# Parse XML to JSON (no code generation)
+informatica-python workflow.xml --json
+
+# Save parsed JSON to file
+informatica-python workflow.xml --json-file parsed.json
+```
+
+### Python API
+
+```python
+from informatica_python import InformaticaConverter
+
+# Convert XML to Python files
+converter = InformaticaConverter(data_lib="pandas")
+converter.convert("workflow.xml", output_dir="output")
+
+# Convert to zip
+converter.convert("workflow.xml", output_zip="output.zip")
+
+# Parse XML to JSON dict
+result = converter.parse_file("workflow.xml")
+
+# Parse XML string
+result = converter.parse_string(xml_string)
+```
+
+## Generated Output Files
+
+| File | Description |
+|------|-------------|
+| `helper_functions.py` | Database/file I/O functions plus Python equivalents for 50+ Informatica expression functions |
+| `mapping_N.py` | One file per mapping with full transformation logic |
+| `workflow.py` | Task orchestration with topological ordering |
+| `config.yml` | Connection configs, source/target metadata, variables |
+| `all_sql_queries.sql` | All extracted SQL queries (source qualifiers, lookups, pre/post SQL) |
+| `error_log.txt` | Conversion summary, warnings, and coverage statistics |
+
+## Supported Transformation Types
+
+- Source Qualifier / Application Source Qualifier
+- Expression
+- Filter
+- Aggregator
+- Sorter
+- Joiner
+- Lookup Procedure
+- Router
+- Union
+- Update Strategy
+- Sequence Generator
+- Normalizer
+- Rank
+- Stored Procedure (placeholder)
+- Custom Transformation (placeholder)
+- Java Transformation (placeholder)
+- SQL Transformation
+
+## Supported Data Libraries
+
+Choose your preferred data manipulation library with `--data-lib`:
+
+- **pandas** (default) — Standard Python data analysis
+- **dask** — Parallel computing with pandas-like API
+- **polars** — Fast DataFrame library written in Rust
+- **vaex** — Out-of-core DataFrames for large datasets
+- **modin** — Drop-in pandas replacement with parallel execution
+
+## Informatica Expression Functions
+
+The generated `helper_functions.py` includes Python equivalents for:
+
+`IIF`, `DECODE`, `NVL`, `NVL2`, `ISNULL`, `LTRIM`, `RTRIM`, `UPPER`, `LOWER`, `SUBSTR`, `LPAD`, `RPAD`, `TO_CHAR`, `TO_DATE`, `TO_INTEGER`, `TO_BIGINT`, `TO_FLOAT`, `TO_DECIMAL`, `REPLACECHR`, `REPLACESTR`, `INSTR`, `LENGTH`, `CONCAT`, `REG_EXTRACT`, `REG_MATCH`, `REG_REPLACE`, `GET_DATE_PART`, `ADD_TO_DATE`, `IS_DATE`, `IS_NUMBER`, `IS_SPACES`, `SYSDATE`, `ERROR`, `ABORT`, and more.
+
+## Requirements
+
+- Python >= 3.8
+- lxml >= 4.9.0
+- PyYAML >= 6.0
+
+## License
+
+MIT

informatica_python-1.0.0.dist-info/RECORD

@@ -0,0 +1,20 @@
+informatica_python/__init__.py,sha256=r3DaPWzFarm6s91zz_VbU1rT20KK-XqXs1a1p6hpk-s,120
+informatica_python/cli.py,sha256=79KpI8p29aATLzi5qH7H2YhMl9I8rzjwMEGzbV48UjM,2480
+informatica_python/converter.py,sha256=6wFPmLc3JyZCgKRYFGYtdD-9SQ4_brkYdB9KIxS1GfY,10506
+informatica_python/models.py,sha256=vZhWYVTKhYC9aqaO3YJFcczFS_mHbpncRs0ME2iLUoI,7190
+informatica_python/parser.py,sha256=ZRcf9x0GV7WbDh5OK_CC2DpswBLIjUcvOMxlYEDcFU0,19756
+informatica_python/generators/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+informatica_python/generators/config_gen.py,sha256=4tqcNKTB06kyGZIiM4yl0q97q_i3zeCHXTjuE1dNFKY,5726
+informatica_python/generators/error_log_gen.py,sha256=98cj2T0Hi3tdH6wQfFI5956QXpqeK11_eND9hzAKm2U,5685
+informatica_python/generators/helper_gen.py,sha256=0i1VwAMZJwzEQWu_AXpWFVbGiGks7jMqp5eTGDZiYv8,34447
+informatica_python/generators/mapping_gen.py,sha256=AEvA_AvoD9iQl9XPE6FuqobwUI5XjEXAboVEnegLfcY,27054
+informatica_python/generators/sql_gen.py,sha256=rwDy-sFpcPZoetUSppK7iF02aFxYIX8PLICnK021o6E,5711
+informatica_python/generators/workflow_gen.py,sha256=ltpDgQELPsERfqSIz1LQUFw_gs-wKqDTOMwb0IDxJpI,9402
+informatica_python/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+informatica_python/utils/datatype_map.py,sha256=iLOYg-iBKT4rMecGbrFkTpJj4yqs5S9HeBOTLUIWhX0,2809
+informatica_python/utils/expression_converter.py,sha256=pmTaY1R5KDVWtJiVD-8PgYIwas_rjKZ_SFhGvn91q2I,3451
+informatica_python-1.0.0.dist-info/METADATA,sha256=LzB1RkDDa1uHI-slVt9mhlxDE82lyWDl-sy93KPP1PE,3344
+informatica_python-1.0.0.dist-info/WHEEL,sha256=PovZm1ExVWmrRefZoXCfejlbKLnQI5SVIf1SWRV4QQI,97
+informatica_python-1.0.0.dist-info/entry_points.txt,sha256=030jjTrx-1oRRQ16HZz52rdcKS8R8_llnymsTUtn_Xc,67
+informatica_python-1.0.0.dist-info/top_level.txt,sha256=Dngg-WNteYi22XAJU2XKAQS8aZ52yM2LYC0tzxrlbVQ,19
+informatica_python-1.0.0.dist-info/RECORD,,

informatica_python-1.0.0.dist-info/WHEEL

@@ -0,0 +1,5 @@
+Wheel-Version: 1.0
+Generator: setuptools (80.9.0.post0)
+Root-Is-Purelib: true
+Tag: py3-none-any
+

informatica_python-1.0.0.dist-info/entry_points.txt

@@ -0,0 +1,2 @@
+[console_scripts]
+informatica-python = informatica_python.cli:main

informatica_python-1.0.0.dist-info/top_level.txt

@@ -0,0 +1 @@
+informatica_python