Update modules install to support version control

.github/workflows/create-lint-wf.yml

@@ -53,7 +53,7 @@ jobs:
         run: nf-core --log-file log.txt lint nf-core-testpipeline --fail-ignored --release
 
       - name: nf-core modules install
-        run: nf-core --log-file log.txt modules install nf-core-testpipeline/ --tool fastqc
+        run: nf-core --log-file log.txt modules install nf-core-testpipeline/ --tool fastqc --latest
 
       - name: Upload log file artifact
         if: ${{ always() }}

CHANGELOG.md

@@ -17,7 +17,7 @@
 
 * Update comment style of modules `functions.nf` template file [[#1076](https://github.com/nf-core/tools/issues/1076)]
 * Use Biocontainers API instead of quayi.io API for `nf-core modules create` [[#875](https://github.com/nf-core/tools/issues/875)]
-
+* Update `nf-core modules install` to handle different versions of modules  [#1116](https://github.com/nf-core/tools/pull/1116)
 #### Sync
 
 * Don't set the default value to `"null"` when a parameter is initialised as `null` in the config [[#1074](https://github.com/nf-core/tools/pull/1074)]

nf_core/__main__.py

@@ -380,7 +380,8 @@ def list(ctx, pipeline_dir, json):
 @click.pass_context
 @click.argument("pipeline_dir", type=click.Path(exists=True), required=True, metavar="<pipeline directory>")
 @click.option("-t", "--tool", type=str, metavar="<tool> or <tool/subtool>")
-def install(ctx, pipeline_dir, tool):
+@click.option("-l", "--latest", is_flag=True, default=False, help="Install the latest version of the module")
+def install(ctx, pipeline_dir, tool, latest):
     """
     Add a DSL2 software wrapper module to a pipeline.
 
@@ -391,7 +392,7 @@ def install(ctx, pipeline_dir, tool):
         mods = nf_core.modules.PipelineModules()
         mods.modules_repo = ctx.obj["modules_repo_obj"]
         mods.pipeline_dir = pipeline_dir
-        mods.install(tool)
+        mods.install(tool, latest)
     except UserWarning as e:
         log.critical(e)
         sys.exit(1)

nf_core/modules/module_utils.py

@@ -0,0 +1,180 @@
+import glob
+import json
+import os
+import requests
+import sys
+import logging
+import questionary
+from itertools import count
+
+from requests import api
+
+import nf_core.utils
+
+from .modules_repo import ModulesRepo
+
+log = logging.getLogger(__name__)
+
+
+def get_module_git_log(module_name, per_page=30, page_nbr=1, since="2020-11-25T00:00:00Z"):
+    """
+    Fetches the commit history the of requested module
+    Args:
+        module_name (str): Name of module
+        per_page (int): Number of commits per page returned by API
+        page_nbr (int): Page number of the retrieved commits
+        since (str): Only show commits later than this timestamp.
+        Time should be given in ISO-8601 format: YYYY-MM-DDTHH:MM:SSZ.
+
+    Returns:
+        [ dict ]: List of commit SHAs and associated (truncated) message
+    """
+    api_url = f"https://api.github.com/repos/nf-core/modules/commits?sha=master&path=software/{module_name}&per_page={per_page}&page={page_nbr}&since={since}"
+    response = requests.get(api_url, auth=nf_core.utils.github_api_auto_auth())
+    if response.status_code == 200:
+        commits = response.json()
+
+        if len(commits) == 0:
+            log.debug(f"Reached end of commit history for '{module_name}'")
+            return []
+        else:
+            # Return the commit SHAs and the first line of the commit message
+            return [
+                {"git_sha": commit["sha"], "trunc_message": commit["commit"]["message"].partition("\n")[0]}
+                for commit in commits
+            ]
+    elif response.status_code == 404:
+        log.error(f"Module '{module_name}' not found in 'nf-core/modules/'\n{api_url}")
+        sys.exit(1)
+    else:
+        raise SystemError(f"Unable to fetch commit SHA for module {module_name}")
+
+
+def create_modules_json(pipeline_dir):
+    """
+    Create the modules.json files
+
+    Args:
+        pipeline_dir (str): The directory where the `modules.json` should be created
+    """
+    pipeline_config = nf_core.utils.fetch_wf_config(pipeline_dir)
+    pipeline_name = pipeline_config["manifest.name"]
+    pipeline_url = pipeline_config["manifest.homePage"]
+    modules_json = {"name": pipeline_name.strip("'"), "homePage": pipeline_url.strip("'"), "modules": {}}
+    all_module_file_paths = glob.glob(f"{pipeline_dir}/modules/nf-core/software/**/*", recursive=True)
+
+    # Extract the module paths from the file paths
+    module_paths = list(set(map(os.path.dirname, filter(os.path.isfile, all_module_file_paths))))
+    module_names = [path.replace(f"{pipeline_dir}/modules/nf-core/software/", "") for path in module_paths]
+    module_repo = ModulesRepo()
+    for module_name, module_path in zip(module_names, module_paths):
+        try:
+            # Find the correct commit SHA for the local files.
+            # We iterate over the commit log pages until we either
+            # find a matching commit or we reach the end of the commits
+            correct_commit_sha = None
+            commit_page_nbr = 1
+            while correct_commit_sha is None:
+
+                commit_shas = [
+                    commit["git_sha"] for commit in get_module_git_log(module_name, page_nbr=commit_page_nbr)
+                ]
+                correct_commit_sha = find_correct_commit_sha(module_name, module_path, module_repo, commit_shas)
+                commit_page_nbr += 1
+
+            modules_json["modules"][module_name] = {"git_sha": correct_commit_sha}
+        except SystemError as e:
+            log.error(e)
+            log.error("Will not create 'modules.json' file")
+            sys.exit(1)
+    modules_json_path = os.path.join(pipeline_dir, "modules.json")
+    with open(modules_json_path, "w") as fh:
+        json.dump(modules_json, fh, indent=4)
+
+
+def find_correct_commit_sha(module_name, module_path, modules_repo, commit_shas):
+    """
+    Returns the SHA for the latest commit where the local files are identical to the remote files
+    Args:
+        module_name (str): Name of module
+        module_path (str): Path to module in local repo
+        module_repo (str): Remote repo for module
+        commit_shas ([ str ]): List of commit SHAs for module, sorted in descending order
+    Returns:
+        commit_sha (str): The latest commit SHA where local files are identical to remote files
+    """
+
+    files_to_check = ["main.nf", "functions.nf", "meta.yml"]
+    local_file_contents = [None, None, None]
+    for i, file in enumerate(files_to_check):
+        try:
+            local_file_contents[i] = open(os.path.join(module_path, file), "r").read()
+        except FileNotFoundError as e:
+            log.debug(f"Could not open file: {os.path.join(module_path, file)}")
+            continue
+    for commit_sha in commit_shas:
+        if local_module_equal_to_commit(local_file_contents, module_name, modules_repo, commit_sha):
+            return commit_sha
+    return None
+
+
+def local_module_equal_to_commit(local_files, module_name, modules_repo, commit_sha):
+    """
+    Compares the local module files to the module files for the given commit sha
+    Args:
+        local_files ([ str ]): Contents of local files. `None` if files doesn't exist
+        module_name (str): Name of module
+        module_repo (str): Remote repo for module
+        commit_sha (str): Commit SHA for remote version to compare against local version
+    Returns:
+        bool: Whether all local files are identical to remote version
+    """
+
+    files_to_check = ["main.nf", "functions.nf", "meta.yml"]
+    files_are_equal = [False, False, False]
+    remote_copies = [None, None, None]
+
+    module_base_url = f"https://raw.githubusercontent.com/{modules_repo.name}/{commit_sha}/software/{module_name}"
+    for i, file in enumerate(files_to_check):
+        # Download remote copy and compare
+        api_url = f"{module_base_url}/{file}"
+        r = requests.get(url=api_url)
+        if r.status_code != 200:
+            log.debug(f"Could not download remote copy of file module {module_name}/{file}")
+            log.debug(api_url)
+        else:
+            try:
+                remote_copies[i] = r.content.decode("utf-8")
+            except UnicodeDecodeError as e:
+                log.debug(f"Could not decode remote copy of {file} for the {module_name} module")
+
+        # Compare the contents of the files.
+        # If the file is missing from both the local and remote repo
+        # we will get the comparision None == None
+        if local_files[i] == remote_copies[i]:
+            files_are_equal[i] = True
+
+    return all(files_are_equal)
+
+
+def prompt_module_version_sha(module):
+    older_commits_choice = questionary.Choice(
+        title=[("fg:ansiyellow", "older commits"), ("class:choice-default", "")], value=""
+    )
+    git_sha = ""
+    page_nbr = 1
+    next_page_commits = get_module_git_log(module, per_page=10, page_nbr=page_nbr)
+    while git_sha is "":
+        commits = next_page_commits
+        next_page_commits = get_module_git_log(module, per_page=10, page_nbr=page_nbr + 1)
+        choices = []
+        for title, sha in map(lambda commit: (commit["trunc_message"], commit["git_sha"]), commits):
+            tag_display = [("fg:ansiblue", f"{title}  "), ("class:choice-default", "")]
+            choices.append(questionary.Choice(title=tag_display, value=sha))
+        if len(next_page_commits) > 0:
+            choices += [older_commits_choice]
+        git_sha = questionary.select(
+            f"Select '{module}' version", choices=choices, style=nf_core.utils.nfcore_question_style
+        ).unsafe_ask()
+        page_nbr += 1
+    return git_sha

nf_core/modules/modules_repo.py

@@ -0,0 +1,114 @@
+import os
+import requests
+import base64
+import sys
+import logging
+import nf_core.utils
+
+log = logging.getLogger(__name__)
+
+
+class ModulesRepo(object):
+    """
+    An object to store details about the repository being used for modules.
+
+    Used by the `nf-core modules` top-level command with -r and -b flags,
+    so that this can be used in the same way by all sucommands.
+    """
+
+    def __init__(self, repo="nf-core/modules", branch="master"):
+        self.name = repo
+        self.branch = branch
+        self.modules_file_tree = {}
+        self.modules_current_hash = None
+        self.modules_avail_module_names = []
+
+    def get_modules_file_tree(self):
+        """
+        Fetch the file list from the repo, using the GitHub API
+
+        Sets self.modules_file_tree
+             self.modules_current_hash
+             self.modules_avail_module_names
+        """
+        api_url = "https://api.github.com/repos/{}/git/trees/{}?recursive=1".format(self.name, self.branch)
+        r = requests.get(api_url, auth=nf_core.utils.github_api_auto_auth())
+        if r.status_code == 404:
+            log.error("Repository / branch not found: {} ({})\n{}".format(self.name, self.branch, api_url))
+            sys.exit(1)
+        elif r.status_code != 200:
+            raise SystemError(
+                "Could not fetch {} ({}) tree: {}\n{}".format(self.name, self.branch, r.status_code, api_url)
+            )
+
+        result = r.json()
+        assert result["truncated"] == False
+
+        self.modules_current_hash = result["sha"]
+        self.modules_file_tree = result["tree"]
+        for f in result["tree"]:
+            if f["path"].startswith("software/") and f["path"].endswith("/main.nf") and "/test/" not in f["path"]:
+                # remove software/ and /main.nf
+                self.modules_avail_module_names.append(f["path"][9:-8])
+
+    def get_module_file_urls(self, module, commit=""):
+        """Fetch list of URLs for a specific module
+
+        Takes the name of a module and iterates over the GitHub repo file tree.
+        Loops over items that are prefixed with the path 'software/<module_name>' and ignores
+        anything that's not a blob. Also ignores the test/ subfolder.
+
+        Returns a dictionary with keys as filenames and values as GitHub API URLs.
+        These can be used to then download file contents.
+
+        Args:
+            module (string): Name of module for which to fetch a set of URLs
+
+        Returns:
+            dict: Set of files and associated URLs as follows:
+
+            {
+                'software/fastqc/main.nf': 'https://api.github.com/repos/nf-core/modules/git/blobs/65ba598119206a2b851b86a9b5880b5476e263c3',
+                'software/fastqc/meta.yml': 'https://api.github.com/repos/nf-core/modules/git/blobs/0d5afc23ba44d44a805c35902febc0a382b17651'
+            }
+        """
+        results = {}
+        for f in self.modules_file_tree:
+            if not f["path"].startswith("software/{}".format(module)):
+                continue
+            if f["type"] != "blob":
+                continue
+            if "/test/" in f["path"]:
+                continue
+            results[f["path"]] = f["url"]
+        if commit != "":
+            for path in results:
+                results[path] = f"https://api.github.com/repos/nf-core/modules/contents/{path}?ref={commit}"
+        return results
+
+    def download_gh_file(self, dl_filename, api_url):
+        """Download a file from GitHub using the GitHub API
+
+        Args:
+            dl_filename (string): Path to save file to
+            api_url (string): GitHub API URL for file
+
+        Raises:
+            If a problem, raises an error
+        """
+
+        # Make target directory if it doesn't already exist
+        dl_directory = os.path.dirname(dl_filename)
+        if not os.path.exists(dl_directory):
+            os.makedirs(dl_directory)
+
+        # Call the GitHub API
+        r = requests.get(api_url, auth=nf_core.utils.github_api_auto_auth())
+        if r.status_code != 200:
+            raise SystemError("Could not fetch {} file: {}\n {}".format(self.name, r.status_code, api_url))
+        result = r.json()
+        file_contents = base64.b64decode(result["content"])
+
+        # Write the file contents
+        with open(dl_filename, "wb") as fh:
+            fh.write(file_contents)