From 15b3fc1b6a64eba925e0775378a917737fcab169 Mon Sep 17 00:00:00 2001
From: Cody <cody@dkdc.dev>
Date: Mon, 16 Sep 2024 22:36:24 -0400
Subject: [PATCH] bug fixes, improved docs etl, docs eda, bump version

---
 eda.qmd                             | 103 ++++++++++++++++++++++++++++
 pyproject.toml                      |   2 +-
 src/ibis_analytics/etl/transform.py |   2 +
 src/ibis_analytics/metrics.py       |   4 +-
 4 files changed, 108 insertions(+), 3 deletions(-)

diff --git a/eda.qmd b/eda.qmd
index 04b2b14..65a69dd 100644
--- a/eda.qmd
+++ b/eda.qmd
@@ -34,6 +34,109 @@ ibis.options.repr.interactive.max_columns = None
 
 px.defaults.template = "plotly_dark"
 ```
+
+```{python}
+def lookback(t, days=None):
+    if days is None:
+        return t
+    return t.filter(t["timestamp"] > ibis.now() - ibis.interval(days=days))
+```
+
+```{python}
+t = docs_t
+t
+```
+
+```{python}
+t.filter(t["path"].startswith("/posts/"))
+```
+
+```{python}
+px.bar(
+    t.filter(t["path"].startswith("/posts/"))
+    .mutate(path=ibis._["path"].re_extract(r"^/posts/[^/]+", 0))
+    # .filter(t["path"].contains("ibis-version-"))
+    .group_by("path")
+    .agg(count=ibis._.count())
+    .order_by(ibis.desc("count"))
+    .limit(10),
+    x="path",
+    y="count",
+    log_y=True,
+)
+```
+
+```{python}
+c = px.line(
+    metrics.docs_rolling_by_path(
+        t.filter(t["path"].startswith("/posts/"))
+        .filter(~ibis._["path"].contains("ibis-version-"))
+        .mutate(path=ibis._["path"].re_extract(r"^/posts/[^/]+", 0)),
+        days=28,
+    ).pipe(lookback, days=None),
+    x="timestamp",
+    y="rolling_docs",
+    color="path",
+    log_y=True,
+)
+# no legend
+c.update_layout(showlegend=False)
+```
+
+
+---
+
+```{python}
+```
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
 
 
 ```{python}
diff --git a/pyproject.toml b/pyproject.toml
index 4395293..f1ea548 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -7,7 +7,7 @@ allow-direct-references = true
 
 [project]
 name = "ibis-analytics"
-version = "0.10.0"
+version = "0.11.0"
 authors = [{ name = "Cody", email = "cody@dkdc.dev" }]
 description = "Ibis analytics with Ibis"
 readme = "readme.md"
diff --git a/src/ibis_analytics/etl/transform.py b/src/ibis_analytics/etl/transform.py
index aeaa994..493d971 100644
--- a/src/ibis_analytics/etl/transform.py
+++ b/src/ibis_analytics/etl/transform.py
@@ -167,6 +167,8 @@ def docs(t):
 
     def transform(t):
         t = t.rename({"path": "2_path", "timestamp": "date"})
+        t = t.mutate(path=ibis._["path"].replace("/index.html", ""))
+        t = t.mutate(path=ibis._["path"].re_extract(r"^/posts/[^/]+", 0))
         return t
 
     docs = t.pipe(preprocess).pipe(transform).pipe(postprocess)
diff --git a/src/ibis_analytics/metrics.py b/src/ibis_analytics/metrics.py
index d263d1c..b62d8b8 100644
--- a/src/ibis_analytics/metrics.py
+++ b/src/ibis_analytics/metrics.py
@@ -92,7 +92,7 @@ def docs_rolling(t: ibis.Table, days: int = 28) -> ibis.Table:
         .over(
             ibis.window(
                 order_by="timestamp",
-                preceding=28,
+                preceding=days,
                 following=0,
             )
         ),
@@ -115,7 +115,7 @@ def docs_rolling_by_path(t: ibis.Table, days: int = 28) -> ibis.Table:
             ibis.window(
                 order_by="timestamp",
                 group_by="path",
-                preceding=28,
+                preceding=days,
                 following=0,
             )
         ),