Implement exercise T6L2/branch-compare

branch_compare/.gitmastery-exercise.json

@@ -0,0 +1,16 @@
+{
+  "exercise_name": "branch-compare",
+  "tags": ["git-branch", "git-diff"],
+  "requires_git": true,
+  "requires_github": false,
+  "base_files": {
+    "answers.txt": "answers.txt"
+  },
+  "exercise_repo": {
+    "repo_type": "local",
+    "repo_name": "data_streams",
+    "repo_title": null,
+    "create_fork": null,
+    "init": true
+  }
+}

branch_compare/README.md

@@ -0,0 +1,9 @@
+# branch-compare
+
+## Task
+
+You are recording a numerical data stream from two sources. The data are stored in `data.txt`, using a different branch for each stream. The two data streams are supposed to be identical but can vary on rare occasions.
+
+Answer the questions given in `answers.txt`.
+
+Run `gitmastery verify` to check if your answers are correct.

branch_compare/download.py

@@ -0,0 +1,62 @@
+from exercise_utils.cli import run_command
+from exercise_utils.git import add, commit, checkout
+from exercise_utils.file import append_to_file
+from exercise_utils.gitmastery import create_start_tag
+
+import random
+
+__resources__ = {}
+
+def get_sequence(n=1000, digits=8, seed=None):
+    rng = random.Random(seed)
+    lo, hi = 10**(digits - 1), 10**digits - 1
+    return rng.sample(range(lo, hi + 1), k=n)
+
+def get_modified_sequence(seq, digits=8, idx=None, seed=None):
+    rng = random.Random(seed)
+    n = len(seq)
+    if idx is None:
+        idx = rng.randrange(n)
+
+    modified = seq.copy()
+    seen = set(seq)
+    lo, hi = 10**(digits - 1), 10**digits - 1
+
+    old = modified[idx]
+    new = old
+    while new in seen:
+        new = rng.randint(lo, hi)
+    modified[idx] = new
+    return modified
+
+
+def setup(verbose: bool = False):
+
+    create_start_tag(verbose)
+
+    orig_data = get_sequence()
+    modified_data = get_modified_sequence(orig_data)
+
+    run_command(["touch", "data.txt"], verbose)
+    add(["data.txt"], verbose)
+    commit("Add empty data.txt", verbose)
+    checkout("stream-1", True, verbose)
+
+    for i in orig_data:
+        append_to_file("data.txt", str(i)+"\n")
+
+    add(["data.txt"], verbose)
+    commit("Add data to data.txt", verbose)
+
+
+    checkout("main", False, verbose)
+    checkout("stream-2", True, verbose)
+
+    for i in modified_data:
+        append_to_file("data.txt", str(i)+"\n")
+
+    add(["data.txt"], verbose)
+    commit("Add data to data.txt", verbose)
+
+    checkout("main", False, verbose)
+

branch_compare/res/answers.txt

@@ -0,0 +1,5 @@
+Q: Which number (write only one number) is present in branch stream-1 but not in branch stream-2?
+A: 
+
+Q: Which number (write only one number) is present in branch stream-2 but not in branch stream-1?
+A: 

branch_compare/tests/specs/base.yml

@@ -0,0 +1,44 @@
+initialization:
+  steps:
+    - type: commit
+      empty: true
+      message: Empty commit
+      id: start
+
+    # Create stream-1 with data.txt containing a unique number 12345
+    - type: branch
+      branch-name: stream-1
+    - type: new-file
+      filename: data.txt
+      contents: |
+        11111
+        22222
+        12345
+    - type: add
+      files:
+        - data.txt
+    - type: commit
+      message: Add data.txt on stream-1
+
+    # Return to main
+    - type: checkout
+      branch-name: main
+
+    # Create stream-2 with data.txt containing a unique number 98765
+    - type: branch
+      branch-name: stream-2
+    - type: new-file
+      filename: data.txt
+      contents: |
+        11111
+        22222
+        98765
+    - type: add
+      files:
+        - data.txt
+    - type: commit
+      message: Add data.txt on stream-2
+
+    # Return to main at the end for verify to start cleanly
+    - type: checkout
+      branch-name: main

branch_compare/tests/specs/extra_commit_on_stream1.yml

@@ -0,0 +1,55 @@
+initialization:
+  steps:
+    - type: commit
+      empty: true
+      message: Empty commit
+      id: start
+
+    # Base setup identical to base.yml
+    - type: branch
+      branch-name: stream-1
+    - type: new-file
+      filename: data.txt
+      contents: |
+        11111
+        22222
+        12345
+    - type: add
+      files:
+        - data.txt
+    - type: commit
+      message: Add data.txt on stream-1
+
+    - type: checkout
+      branch-name: main
+
+    - type: branch
+      branch-name: stream-2
+    - type: new-file
+      filename: data.txt
+      contents: |
+        11111
+        22222
+        98765
+    - type: add
+      files:
+        - data.txt
+    - type: commit
+      message: Add data.txt on stream-2
+
+    # Make an extra change on stream-1 to simulate user changes
+    - type: checkout
+      branch-name: stream-1
+    - type: new-file
+      filename: extra.txt
+      contents: |
+        extra content
+    - type: add
+      files:
+        - extra.txt
+    - type: commit
+      message: Extra change on stream-1
+
+    # Return to main at the end
+    - type: checkout
+      branch-name: main

branch_compare/tests/test_verify.py

@@ -0,0 +1,55 @@
+from git_autograder import GitAutograderStatus, GitAutograderTestLoader, assert_output
+from git_autograder.answers.rules.has_exact_value_rule import HasExactValueRule
+
+from ..verify import verify, QUESTION_ONE, QUESTION_TWO, NO_CHANGES_ERROR
+
+REPOSITORY_NAME = "branch-compare"
+
+loader = GitAutograderTestLoader(__file__, REPOSITORY_NAME, verify)
+
+
+def test_base():
+    with loader.load(
+        "specs/base.yml",
+        "start",
+        mock_answers={
+            QUESTION_ONE: "12345",
+            QUESTION_TWO: "98765",
+        },
+    ) as output:
+        assert_output(output, GitAutograderStatus.SUCCESSFUL)
+
+def test_wrong_stream1_diff():
+    with loader.load(
+        "specs/base.yml",
+        "start",
+        mock_answers={
+            QUESTION_ONE: "99999",
+            QUESTION_TWO: "98765",
+        },
+    ) as output:
+        assert_output(
+            output,
+            GitAutograderStatus.UNSUCCESSFUL,
+            [HasExactValueRule.NOT_EXACT.format(question=QUESTION_ONE)],
+        )
+
+def test_wrong_stream2_diff():
+    with loader.load(
+        "specs/base.yml",
+        "start",
+        mock_answers={
+            QUESTION_ONE: "12345",
+            QUESTION_TWO: "99999",
+        },
+    ) as output:
+        assert_output(
+            output,
+            GitAutograderStatus.UNSUCCESSFUL,
+            [HasExactValueRule.NOT_EXACT.format(question=QUESTION_TWO)],
+        )
+
+
+def test_changes_made_extra_commit():
+    with loader.load("specs/extra_commit_on_stream1.yml", "start") as output:
+        assert_output(output, GitAutograderStatus.UNSUCCESSFUL, [NO_CHANGES_ERROR])

branch_compare/verify.py

@@ -0,0 +1,83 @@
+from git_autograder import (
+    GitAutograderOutput,
+    GitAutograderExercise,
+    GitAutograderStatus,
+)
+
+from git_autograder.answers.rules import HasExactValueRule, NotEmptyRule
+
+
+QUESTION_ONE = "Which number (write only one number) is present in branch stream-1 but not in branch stream-2?"
+QUESTION_TWO = "Which number (write only one number) is present in branch stream-2 but not in branch stream-1?"
+NO_CHANGES_ERROR = "No changes are supposed to be made to the two branches in this exercise"
+
+FILE_PATH = "data.txt"
+BRANCH_1 = "stream-1"
+BRANCH_2 = "stream-2"
+
+def has_made_changes(exercise: GitAutograderExercise) -> bool:
+    repo = exercise.repo.repo
+
+    for bname in (BRANCH_1, BRANCH_2):
+        if not exercise.repo.branches.has_branch(bname):
+            return True
+
+        head = repo.commit(bname)
+
+        if len(head.parents) != 1:
+            return True
+
+        # Count commits unique to branch relative to main
+        merge_bases = repo.merge_base(bname, "main")
+        if not merge_bases:
+            return True
+        base = merge_bases[0]
+        unique_commits = list(repo.iter_commits(f"{base.hexsha}..{bname}"))
+        if len(unique_commits) != 1:
+            return True
+
+    return False
+
+def get_branch_diff(exercise: GitAutograderExercise, branch1: str, branch2: str) -> str:
+    exercise.repo.branches.branch(branch1).checkout()
+    with exercise.repo.files.file(FILE_PATH) as f1:
+        contents1 = f1.read()
+
+    exercise.repo.branches.branch(branch2).checkout()
+    with exercise.repo.files.file(FILE_PATH) as f2:
+        contents2 = f2.read()
+
+    exercise.repo.branches.branch("main").checkout()
+
+    set1 = {line.strip() for line in contents1.splitlines() if line.strip()}
+    set2 = {line.strip() for line in contents2.splitlines() if line.strip()}
+    diff = set1 - set2
+    return str(diff.pop())
+
+def get_stream1_diff(exercise: GitAutograderExercise) -> str:
+    return get_branch_diff(exercise, BRANCH_1, BRANCH_2)
+
+def get_stream2_diff(exercise: GitAutograderExercise) -> str:
+    return get_branch_diff(exercise, BRANCH_2, BRANCH_1)
+
+def verify(exercise: GitAutograderExercise) -> GitAutograderOutput:
+
+    if has_made_changes(exercise):
+        return exercise.to_output([NO_CHANGES_ERROR], GitAutograderStatus.UNSUCCESSFUL)
+
+    exercise.repo.branches.branch("main").checkout()
+
+    ans_1 = get_stream1_diff(exercise)
+    ans_2 = get_stream2_diff(exercise)
+
+    exercise.answers.add_validation(
+        QUESTION_ONE,
+        NotEmptyRule(),
+        HasExactValueRule(ans_1),
+    ).add_validation(
+        QUESTION_TWO, 
+        NotEmptyRule(),
+        HasExactValueRule(ans_2),
+    ).validate()
+
+    return exercise.to_output(["Great work comparing the branches successfully!"], GitAutograderStatus.SUCCESSFUL)