List system ms

packages/df-wiki-cli/df_wiki_cli/content/main.py

 import typer
 import sys
 import re
+import json
+import pandas as pd
+from pandas.errors import InvalidIndexError
 import shutil
 from typing_extensions import Annotated
 from typing import Optional, List
@@ -111,3 +114,80 @@ def structure(
         systemDir = output / system
         systemDir.mkdir(parents=True, exist_ok=True)
         shutil.copy2(f, systemDir)
+
+
+@app.command()
+def systems(
+    dir: Annotated[
+        Path,
+        typer.Option(exists=False, file_okay=False, readable=True, dir_okay=True),
+    ],
+    pfam: Annotated[
+        Path,
+        typer.Option(
+            exists=False,
+            file_okay=True,
+            writable=True,
+        ),
+    ],
+    output: Annotated[
+        Path,
+        typer.Option(
+            file_okay=True,
+            dir_okay=False,
+            writable=True,
+            resolve_path=True,
+        ),
+    ],
+):
+    with open(pfam, "r") as pf:
+        pfam_df = pd.read_csv(pf, index_col="AC", keep_default_na=False)
+        systems = []
+        with open(output, "a") as ty:
+            for file in dir.iterdir():
+                if file.suffix == ".md":
+                    console.rule(f"[bold blue]{file.name}", style="blue")
+                    with open(file) as f:
+                        metadata, _ = frontmatter.parse(f.read())
+                        del metadata["layout"]
+                        if "tableColumns" in metadata:
+                            table_data = metadata["tableColumns"]
+                            if "PFAM" in table_data:
+                                # print(table_data["PFAM"])
+                                pfams_list = [
+                                    pfam.strip()
+                                    for pfam in table_data["PFAM"].split(",")
+                                ]
+                                pfam_metadata = list()
+                                for pfam in pfams_list:
+                                    try:
+                                        pfam_obj = pfam_df.loc[[pfam]]
+                                        # print(pfam_obj)
+                                        pfam_to_dict = pfam_obj.to_dict(orient="index")
+                                        pfam_to_dict[pfam]["AC"] = pfam
+                                        flatten_value = pfam_to_dict[pfam]
+                                        pfam_metadata.append(flatten_value)
+
+                                    except KeyError as err:
+                                        console.print(f"[bold red]{err}", style="red")
+                                        console.print(
+                                            f"[bold red]No pfam entry or {pfam}",
+                                            style="red",
+                                        )
+                                        continue
+                                del metadata["tableColumns"]
+
+                                if "article" in table_data:
+                                    metadata["doi"] = table_data["article"]["doi"]
+                                    if "abstract" in table_data["article"]:
+                                        metadata["abstract"] = table_data["article"][
+                                            "abstract"
+                                        ]
+                                    del table_data["article"]
+                                sanitizedMetadata = {**metadata, **table_data}
+
+                                sanitizedMetadata["PFAM"] = pfam_metadata
+                                systems.append(sanitizedMetadata)
+
+            json_object = json.dumps(systems, indent=2)
+            ty.write(json_object)