perf: fast-path substring guards + short-circuit ordering fixes in parser matches()

src/calcflow/io/orca/blocks/charges.py

@@ -40,9 +40,15 @@ def matches(self, line: str, state: ParseState) -> bool:
         Check if this line marks the beginning of an atomic charges block.
 
         Returns True for either MULLIKEN or LOEWDIN atomic charges headers.
-        Note: No state flag check - allows parser to run multiple times (once per method).
+        Allows the parser to run twice (once per method); stops after both are parsed.
         """
-        return bool(MULLIKEN_START_PAT.search(line)) or bool(LOEWDIN_START_PAT.search(line))
+        if state.parsed_mulliken and state.parsed_loewdin:
+            return False
+        if "MULLIKEN" not in line and "LOEWDIN" not in line:
+            return False
+        if not state.parsed_mulliken and MULLIKEN_START_PAT.search(line):
+            return True
+        return not state.parsed_loewdin and bool(LOEWDIN_START_PAT.search(line))
 
     def parse(self, iterator: PeekableIterator, start_line: str, state: ParseState) -> None:
         """

src/calcflow/io/orca/blocks/dispersion.py

@@ -51,6 +51,8 @@ def matches(self, line: str, state: ParseState) -> bool:
         """
         Check if this line marks the start of a dispersion block.
         """
+        if "DISPERSION CORRECTION" not in line:
+            return False
         if state.parsed_dispersion:
             return False
         return bool(START_PAT.search(line))

src/calcflow/io/orca/blocks/finalization.py

@@ -9,6 +9,8 @@
 
 class FinalEnergyParser:
     def matches(self, line: str, state: ParseState) -> bool:
+        if "FINAL SINGLE POINT ENERGY" not in line:
+            return False
         return state.final_energy is None and bool(FINAL_ENERGY_PAT.search(line))
 
     def parse(self, iterator: PeekableIterator, start_line: str, state: ParseState) -> None:
@@ -19,6 +21,8 @@ def parse(self, iterator: PeekableIterator, start_line: str, state: ParseState)
 
 class TerminationParser:
     def matches(self, line: str, state: ParseState) -> bool:
+        if "ORCA TERMINATED" not in line:
+            return False
         return state.termination_status == "UNKNOWN" and bool(NORMAL_TERM_PAT.search(line))
 
     def parse(self, iterator: PeekableIterator, start_line: str, state: ParseState) -> None:

src/calcflow/io/orca/blocks/geometry.py

@@ -13,6 +13,8 @@
 
 class GeometryParser:
     def matches(self, line: str, state: ParseState) -> bool:
+        if "CARTESIAN COORDINATES" not in line:
+            return False
         return not state.parsed_geometry and bool(GEOMETRY_START_PAT.search(line))
 
     def parse(self, iterator: PeekableIterator, start_line: str, state: ParseState) -> None:

src/calcflow/io/orca/blocks/orbitals.py

@@ -50,6 +50,8 @@ def matches(self, line: str, state: ParseState) -> bool:
         Returns:
             True if this is the start of an orbital block and hasn't been parsed yet.
         """
+        if "ORBITAL ENERGIES" not in line:
+            return False
         if state.parsed_orbitals:
             return False
         return bool(ORBITAL_ENERGIES_START_PAT.search(line))

src/calcflow/io/orca/blocks/timing.py

@@ -35,6 +35,8 @@ def __init__(self):
 
     def matches(self, line: str, state: ParseState) -> bool:
         """Matches on total wall time line or module timing lines."""
+        if "..." not in line and "TOTAL RUN TIME" not in line:
+            return False
         if state.parsed_timing:
             return False
         return bool(TOTAL_TIME_PAT.search(line)) or bool(MODULE_TIME_PAT.search(line))

src/calcflow/io/qchem/blocks/adc/excited_states.py

@@ -56,6 +56,8 @@ class AdcExcitedStatesParser(BlockParser):
     START_PAT: ClassVar[re.Pattern] = re.compile(r"Excited State Summary")
 
     def matches(self, line: str, state: ParseState) -> bool:
+        if "Excited State Summary" not in line:
+            return False
         if state.parsed_adc_excited:
             return False
         return bool(self.START_PAT.search(line))

src/calcflow/io/qchem/blocks/adc/ground_state.py

@@ -41,6 +41,8 @@ class AdcGroundStateParser(BlockParser):
     DAVIDSON_PAT: ClassVar[re.Pattern] = re.compile(r"Starting Davidson for excited states")
 
     def matches(self, line: str, state: ParseState) -> bool:
+        if "A D C" not in line:
+            return False
         if state.parsed_adc_gs:
             return False
         return bool(self.BANNER_PAT.search(line))

src/calcflow/io/qchem/blocks/charges.py

@@ -40,7 +40,9 @@ def matches(self, line: str, state: ParseState) -> bool:
         Returns True only if we haven't already parsed Mulliken charges and the line
         contains the Mulliken charges header.
         """
-        return bool(MULLIKEN_START_PAT.search(line)) and not state.parsed_mulliken
+        if "Mulliken" not in line:
+            return False
+        return not state.parsed_mulliken and bool(MULLIKEN_START_PAT.search(line))
 
     def parse(self, iterator: PeekableIterator, start_line: str, state: ParseState) -> None:
         """

src/calcflow/io/qchem/blocks/cm5.py

@@ -32,7 +32,9 @@ class Cm5Parser:
     """
 
     def matches(self, line: str, state: ParseState) -> bool:
-        return bool(CM5_START_PAT.search(line)) and not state.parsed_cm5
+        if "Charge Model 5" not in line:
+            return False
+        return not state.parsed_cm5 and bool(CM5_START_PAT.search(line))
 
     def parse(self, iterator: PeekableIterator, start_line: str, state: ParseState) -> None:
         logger.debug("Parsing QChem CM5 charges block.")

src/calcflow/io/qchem/blocks/geometry.py

@@ -34,12 +34,14 @@ class GeometryParser:
 
     def matches(self, line: str, state: ParseState) -> bool:
         """Checks if the line starts either the input or standard geometry block."""
-        if INPUT_GEOM_START_PAT.search(line) and not state.parsed_geometry:
+        if "$molecule" not in line.lower() and "Standard Nuclear Orientation" not in line:
+            return False
+        if not state.parsed_geometry and INPUT_GEOM_START_PAT.search(line):
             # We use a single 'parsed_geometry' flag for the input geometry
             return True
         # The final geometry is part of the main calculation, not a separate block to be parsed once
         # For now, let's assume we only parse the final geometry once.
-        return (STANDARD_GEOM_START_PAT.search(line) is not None) and state.final_geometry is None
+        return state.final_geometry is None and bool(STANDARD_GEOM_START_PAT.search(line))
 
     def parse(self, iterator: PeekableIterator, start_line: str, state: ParseState) -> None:
         """Delegates to the appropriate parsing method based on the start line."""

src/calcflow/io/qchem/blocks/hirshfeld.py

@@ -34,7 +34,9 @@ class HirshfeldParser:
     """
 
     def matches(self, line: str, state: ParseState) -> bool:
-        return bool(HIRSHFELD_START_PAT.search(line)) and not state.parsed_hirshfeld
+        if "Hirshfeld" not in line:
+            return False
+        return not state.parsed_hirshfeld and bool(HIRSHFELD_START_PAT.search(line))
 
     def parse(self, iterator: PeekableIterator, start_line: str, state: ParseState) -> None:
         logger.debug("Parsing QChem Hirshfeld charges block.")

src/calcflow/io/qchem/blocks/metadata.py

@@ -23,7 +23,9 @@ def matches(self, line: str, state: ParseState) -> bool:
         """
         if state.metadata.software_version is not None:
             return False
-        return QCHEM_VERSION_PAT.search(line) is not None
+        if "Q-Chem" not in line:
+            return False
+        return bool(QCHEM_VERSION_PAT.search(line))
 
     def parse(self, iterator: PeekableIterator, start_line: str, state: ParseState) -> None:
         """

src/calcflow/io/qchem/blocks/multipole.py

@@ -60,7 +60,9 @@ class MultipoleParser:
 
     def matches(self, line: str, state: ParseState) -> bool:
         """Check if this line marks the beginning of the multipole moments block."""
-        return bool(MULTIPOLE_START_PAT.search(line)) and not state.parsed_multipole
+        if "Multipole Moments" not in line:
+            return False
+        return not state.parsed_multipole and bool(MULTIPOLE_START_PAT.search(line))
 
     def parse(self, iterator: PeekableIterator, start_line: str, state: ParseState) -> None:
         """

src/calcflow/io/qchem/blocks/orbitals.py

@@ -39,6 +39,8 @@ class OrbitalsParser:
 
     def matches(self, line: str, state: ParseState) -> bool:
         """Check if this line marks the start of the orbital energies block."""
+        if "Orbital Energies" not in line:
+            return False
         if state.parsed_orbitals:
             return False
         return bool(ORBITAL_BLOCK_START_PAT.search(line))

src/calcflow/io/qchem/blocks/scf.py

@@ -33,6 +33,8 @@ class ScfParser:
     """Parses the main SCF block, including iterations, final energies, and SMD summaries."""
 
     def matches(self, line: str, state: ParseState) -> bool:
+        if "General SCF" not in line:
+            return False
         return not state.parsed_scf and bool(SCF_START_PAT.search(line))
 
     def parse(self, iterator: PeekableIterator, start_line: str, state: ParseState) -> None:

src/calcflow/io/qchem/blocks/tddft/excitations.py

@@ -56,6 +56,8 @@ def matches(self, line: str, state: ParseState) -> bool:
 
         Matches either TDA or full TDDFT block start, and only if not already parsed.
         """
+        if "Excitation" not in line:
+            return False
         if state.parsed_tddft_tda and state.parsed_tddft_full:
             return False
 

src/calcflow/io/qchem/blocks/tddft/gs_ref.py

@@ -65,6 +65,8 @@ def matches(self, line: str, state: ParseState) -> bool:
 
         Matches "Ground State (Reference) :" and only if not already parsed.
         """
+        if "Ground State" not in line:
+            return False
         if state.parsed_tddft_gs_ref:
             return False
 

src/calcflow/io/qchem/blocks/tddft/nto.py

@@ -54,6 +54,8 @@ def matches(self, line: str, state: ParseState) -> bool:
 
         Returns False if already parsed.
         """
+        if "NTO" not in line:
+            return False
         if state.parsed_nto:
             return False
 

src/calcflow/io/qchem/blocks/tddft/trans_dm.py

@@ -46,6 +46,8 @@ class TransitionDensityMatrixParser(BlockParser):
     END_PAT: ClassVar[re.Pattern] = re.compile(r"^\s*-{5,}\s*$")
 
     def matches(self, line: str, state: ParseState) -> bool:
+        if "Transition Density" not in line:
+            return False
         if state.parsed_tddft_trans_dm:
             return False
         return bool(self.START_PAT.search(line))

src/calcflow/io/qchem/blocks/tddft/unrel_dm.py

@@ -46,6 +46,8 @@ class UnrelaxedDensityMatrixParser(BlockParser):
     END_PAT: ClassVar[re.Pattern] = re.compile(r"^\s*-{5,}\s*$")
 
     def matches(self, line: str, state: ParseState) -> bool:
+        if "Unrelaxed" not in line:
+            return False
         if state.parsed_tddft_unrelaxed_dm:
             return False
         return bool(self.START_PAT.search(line))

src/calcflow/io/qchem/blocks/timing.py

@@ -23,6 +23,8 @@ class TimingParser:
 
     def matches(self, line: str, state: ParseState) -> bool:
         """Matches on the total job time line."""
+        if "Total job time" not in line:
+            return False
         if state.parsed_timing:
             return False
         return bool(TOTAL_TIME_PAT.search(line))