Fix reasoned amendments

src/parl_motion_detector/agreements.py

@@ -107,15 +107,18 @@ def construct_motion(self):
         else:
             motion_lines = [self.preceeding_text, self.agreed_text]
 
-        return Motion(
+        motion = Motion(
             date=self.date,
             chamber=self.chamber,
             major_heading_id=self.major_heading_id,
             minor_heading_id=self.minor_heading_id,
             speech_id=self.speech_id,
             speech_start_pid=self.paragraph_pid,
             motion_lines=motion_lines,
-        ).add_title()
+        )
+        motion.add_title()
+        motion.self_flag()
+        return motion
 
     @computed_field
     @property
@@ -165,15 +168,18 @@ def construct_motion(self):
         Sometimes (like for clauses) there isn't actually a perfect motion to hold onto
         We're just going to cheat and make one.
         """
-        return Motion(
+        motion = Motion(
             date=self.date,
             chamber=self.chamber,
             major_heading_id=self.major_heading_id,
             minor_heading_id=self.minor_heading_id,
             speech_id=self.speech_id,
             speech_start_pid="",
             motion_lines=[self.minor_heading_text, self.preceding_speech],
-        ).add_title()
+        )
+        motion.add_title()
+        motion.self_flag()
+        return motion
 
     @computed_field
     @property

src/parl_motion_detector/mapper.py

@@ -102,6 +102,7 @@ def get_manual_text(data_dir: Path) -> dict[str, Motion]:
     ]
 )
 
+
 preamble = ["Motion made, and Question put,", "Resolved,"]
 
 similar_phrases = {"additional amendment": "amendment"}
@@ -399,6 +400,27 @@ def multiple_decision_assignment(
         possible_motions = condense_motions(possible_motions)
         previous_loop = len(decisions) + 1
 
+        second_stage_motions = [
+            x for x in possible_motions if x.has_flag(Flag.SECOND_STAGE)
+        ]
+        if len(second_stage_motions) > 1:
+            # if we have multiple second stage motions - we want to just keep the first
+            # because it might confuse reasoned amendment assignment
+            possible_motions = [
+                x for x in possible_motions if x not in second_stage_motions[1:]
+            ]
+
+        if any(x.has_flag(Flag.REASONED_AMENDMENT_FULL) for x in possible_motions):
+            # remove any partial reasoned amendments if so
+            # this is because ANNOYINGLY sometimes they don't print the reasoned amendment
+            # so we need to fall back to another way of catagorising it to get the right map
+            # given this is ALWAYS in the same vote as a staged vote
+            possible_motions = [
+                x
+                for x in possible_motions
+                if not x.has_flag(Flag.REASONED_AMENDMENT_PARTIAL)
+            ]
+
         while len(decisions) < previous_loop:
             previous_loop = len(decisions)
             # print(len(decisions), previous_loop)
@@ -428,6 +450,28 @@ def multiple_decision_assignment(
                         raise ValueError("After decision motion not removed")
                     continue
 
+            # hints from flags
+            # if we try and construct a motion from the text surrounding the decision
+            # does it share a flag with one of the motions
+            for decision in decisions:
+                constructed_motion = decision.construct_motion()
+                possible_flagged_motions = []
+                for motion in possible_motions:
+                    overlap_flags = set(motion.flags) & set(constructed_motion.flags)
+                    if overlap_flags:
+                        possible_flagged_motions.append(motion)
+                if len(possible_flagged_motions) == 1:
+                    self.assign_motion_decision(
+                        possible_flagged_motions[0],
+                        decision,
+                        "constructed motion flag match",
+                    )
+                    decisions = [x for x in decisions if x != decision]
+                    possible_motions = [
+                        x for x in possible_motions if x != possible_flagged_motions[0]
+                    ]
+                    continue
+
             if self.chamber == Chamber.SCOTLAND:
                 # Here we have some special casing for amendments in the scottish Parliament
                 # these should happen pretty good in order! But there may be a bit more spacing than usual

src/parl_motion_detector/motions.py

@@ -77,6 +77,9 @@ class Flag(StrEnum):
     MAIN_QUESTION = "main_question"
     MOTION_AMENDMENT = "motion_amendment"
     SCOTTISH_EXPANDED_MOTION = "scottish_expanded_motion"
+    REASONED_AMENDMENT_FULL = "reasoned_amendment_full"
+    REASONED_AMENDMENT_PARTIAL = "reasoned_amendment_partial"
+    SECOND_STAGE = "second_stage"
 
 
 class Motion(BaseModel):
@@ -167,11 +170,18 @@ def self_flag(self):
         """
         Any extra tags to add based on the final content
         """
-        content = str(self).lower()
+        content = str(self).lower().replace("\n", " ").replace("  ", " ")
         if len(self.motion_lines) < 3:
             if abstract_motion(content):
                 self.add_flag(Flag.ABSTRACT_MOTION)
 
+        if reasoned_amendment_full(content):
+            self.add_flag(Flag.REASONED_AMENDMENT_FULL)
+        if second_time_flag(content):
+            self.add_flag(Flag.SECOND_STAGE)
+        if reasoned_amendment_partial(content):
+            self.add_flag(Flag.REASONED_AMENDMENT_PARTIAL)
+
         if amendment_flag(content):
             self.add_flag(Flag.MOTION_AMENDMENT)
         elif main_question(content):
@@ -225,12 +235,23 @@ def dump_test_data(self, tests_data_path: Path):
     ]
 )
 
+reasoned_amendment_partial = PhraseDetector(
+    criteria=[
+        re.compile(r"The reasoned amendment .* has been selected", re.IGNORECASE),
+    ]
+)
+
+reasoned_amendment_full = PhraseDetector(
+    criteria=["declines to give a second reading", "declines to give a third reading"]
+)
+
 # for adding a flag after bringing the contents together
 amendment_flag = PhraseDetector(
     criteria=[
         "I beg to move an amendment",
         "I beg to move amendment",
         "Amendment proposed: at the end of the Question",
+        re.compile(r"The reasoned amendment .* has been selected", re.IGNORECASE),
         re.compile(
             r"question is, (?:that|the) amendment \d+\w? be agreed to\. Are we(?: all)? agreed\?",
             re.IGNORECASE,
@@ -377,6 +398,17 @@ def dump_test_data(self, tests_data_path: Path):
             re.IGNORECASE,
         ),
         re.compile(r"The next question is, that motion .* be agreed to", re.IGNORECASE),
+        re.compile(r"The reasoned amendment .* has been selected", re.IGNORECASE),
+    ]
+)
+
+second_time_flag = PhraseDetector(
+    criteria=[
+        "That the Bill be now read a Second time",
+        "Motion made, That the Bill be read be now read a Second time",
+        "Motion made, That the Bill be now read a Secondtime",
+        "That the Bill be now read a second time",
+        "the Bill be now read a Second time",
     ]
 )
 
@@ -440,6 +472,7 @@ def dump_test_data(self, tests_data_path: Path):
             re.IGNORECASE,
         ),
         re.compile(r"The next question is, that motion .* be agreed to", re.IGNORECASE),
+        re.compile(r"The reasoned amendment .* has been selected", re.IGNORECASE),
     ]
 )