Merge pull request #387 from pbs-data-solutions/parse-native-xml-stirngs

Add functions to parse native xml strings

Author: sanders41

prelude_parser/__init__.py

@@ -1,14 +1,20 @@
 from prelude_parser._prelude_parser import (
     parse_site_native_file,
+    parse_site_native_string,
     parse_subject_native_file,
+    parse_subject_native_string,
     parse_user_native_file,
+    parse_user_native_string,
 )
 from prelude_parser.parser import parse_to_classes, parse_to_dict
 
 __all__ = [
     "parse_site_native_file",
+    "parse_site_native_string",
     "parse_subject_native_file",
+    "parse_subject_native_string",
     "parse_user_native_file",
+    "parse_user_native_string",
     "parse_to_classes",
     "parse_to_dict",
 ]

prelude_parser/_prelude_parser.pyi

@@ -16,8 +16,11 @@ def _parse_flat_file_to_pandas_dict(
     xml_file: str | Path, short_names: bool
 ) -> dict[str, FlatFormInfo]: ...
 def parse_site_native_file(xml_file: str | Path) -> SiteNative: ...
+def parse_site_native_string(xml_str: str) -> SiteNative: ...
 def parse_subject_native_file(xml_file: str | Path) -> SubjectNative: ...
+def parse_subject_native_string(xml_str: str) -> SubjectNative: ...
 def parse_user_native_file(xml_file: str | Path) -> UserNative: ...
+def parse_user_native_string(xml_str: str) -> UserNative: ...
 
 class FileNotFoundError(Exception):
     pass

src/lib.rs

@@ -10,8 +10,11 @@ pub use prelude_xml_parser::native::{
     site_native::SiteNative, subject_native::SubjectNative, user_native::UserNative,
 };
 use prelude_xml_parser::parse_site_native_file as parse_site_native_file_rs;
+use prelude_xml_parser::parse_site_native_string as parse_site_native_string_rs;
 use prelude_xml_parser::parse_subject_native_file as parse_subject_native_file_rs;
+use prelude_xml_parser::parse_subject_native_string as parse_subject_native_string_rs;
 use prelude_xml_parser::parse_user_native_file as parse_user_native_file_rs;
+use prelude_xml_parser::parse_user_native_string as parse_user_native_string_rs;
 use pyo3::prelude::*;
 use pyo3::types::{IntoPyDict, PyDict, PyList};
 use roxmltree::Document;
@@ -244,6 +247,14 @@ fn parse_site_native_file(_py: Python, xml_file: PathBuf) -> PyResult<SiteNative
     }
 }
 
+#[pyfunction]
+fn parse_site_native_string(_py: Python, xml_str: &str) -> PyResult<SiteNative> {
+    match parse_site_native_string_rs(xml_str) {
+        Ok(native) => Ok(native),
+        Err(e) => Err(ParsingError::new_err(format!("Error parsing xml: {:?}", e))),
+    }
+}
+
 #[pyfunction]
 fn parse_subject_native_file(_py: Python, xml_file: PathBuf) -> PyResult<SubjectNative> {
     match parse_subject_native_file_rs(&xml_file) {
@@ -255,6 +266,14 @@ fn parse_subject_native_file(_py: Python, xml_file: PathBuf) -> PyResult<Subject
     }
 }
 
+#[pyfunction]
+fn parse_subject_native_string(_py: Python, xml_str: &str) -> PyResult<SubjectNative> {
+    match parse_subject_native_string_rs(xml_str) {
+        Ok(native) => Ok(native),
+        Err(e) => Err(ParsingError::new_err(format!("Error parsing xml: {:?}", e))),
+    }
+}
+
 #[pyfunction]
 fn parse_user_native_file(_py: Python, xml_file: PathBuf) -> PyResult<UserNative> {
     match parse_user_native_file_rs(&xml_file) {
@@ -266,6 +285,14 @@ fn parse_user_native_file(_py: Python, xml_file: PathBuf) -> PyResult<UserNative
     }
 }
 
+#[pyfunction]
+fn parse_user_native_string(_py: Python, xml_str: &str) -> PyResult<UserNative> {
+    match parse_user_native_string_rs(xml_str) {
+        Ok(native) => Ok(native),
+        Err(e) => Err(ParsingError::new_err(format!("Error parsing xml: {:?}", e))),
+    }
+}
+
 #[pymodule]
 fn _prelude_parser(py: Python, m: &Bound<'_, PyModule>) -> PyResult<()> {
     m.add_class::<SiteNative>()?;
@@ -274,8 +301,11 @@ fn _prelude_parser(py: Python, m: &Bound<'_, PyModule>) -> PyResult<()> {
     m.add_function(wrap_pyfunction!(_parse_flat_file_to_dict, m)?)?;
     m.add_function(wrap_pyfunction!(_parse_flat_file_to_pandas_dict, m)?)?;
     m.add_function(wrap_pyfunction!(parse_site_native_file, m)?)?;
+    m.add_function(wrap_pyfunction!(parse_site_native_string, m)?)?;
     m.add_function(wrap_pyfunction!(parse_subject_native_file, m)?)?;
+    m.add_function(wrap_pyfunction!(parse_subject_native_string, m)?)?;
     m.add_function(wrap_pyfunction!(parse_user_native_file, m)?)?;
+    m.add_function(wrap_pyfunction!(parse_user_native_string, m)?)?;
     m.add(
         "FileNotFoundError",
         py.get_type_bound::<FileNotFoundError>(),

tests/test_parser.py

@@ -4,32 +4,59 @@
 
 from prelude_parser import (
     parse_site_native_file,
+    parse_site_native_string,
     parse_subject_native_file,
+    parse_subject_native_string,
     parse_to_classes,
     parse_to_dict,
     parse_user_native_file,
+    parse_user_native_string,
 )
 from prelude_parser._prelude_parser import FileNotFoundError, InvalidFileTypeError, ParsingError
 
 
-def test_parse_site_native(site_native_xml):
+def test_parse_site_native_file(site_native_xml):
     result = parse_site_native_file(site_native_xml)
 
     assert result.sites[0].name == "Some Site"
 
 
-def test_parse_subject_native(subject_native_xml):
+def test_parse_site_native_string(site_native_xml):
+    with open(site_native_xml) as f:
+        xml = f.read()
+    result = parse_site_native_string(xml)
+
+    assert result.sites[0].name == "Some Site"
+
+
+def test_parse_subject_native_file(subject_native_xml):
     result = parse_subject_native_file(subject_native_xml)
 
     assert result.patients[0].patient_id == "ABC-001"
 
 
-def test_parse_user_native(user_native_xml):
+def test_parse_subject_native_string(subject_native_xml):
+    with open(subject_native_xml) as f:
+        xml = f.read()
+    result = parse_subject_native_string(xml)
+
+    assert result.patients[0].patient_id == "ABC-001"
+
+
+def test_parse_user_native_file(user_native_xml):
     result = parse_user_native_file(user_native_xml)
 
     assert result.users[0].unique_id == "1691421275437"
 
 
+def test_parse_user_native_string(user_native_xml):
+    with open(user_native_xml) as f:
+        xml = f.read()
+    result = parse_user_native_string(xml)
+
+    assert result.users[0].unique_id == "1691421275437"
+
+
 def test_parse_to_classes(test_file_1):
     result = parse_to_classes(test_file_1)
     assert len(result) == 2