slide_status.Rmd

---
title: "Lecture Slide Status Overview"
author: "`r Sys.info()[['user']]`"
date: '`r format(Sys.time(), format = "%F %T", tz = "Europe/Berlin", usetz = TRUE)`'
output:
  html_document:
    toc: yes
    toc_depth: 5
    toc_float:
      collapsed: true
    self_contained: no
    mathjax: null
editor_options:
  chunk_output_type: console
---

```{css, echo=FALSE}
/*
Apply some custom CSS to add breathing room between sub-tables
and we don't bother creating a separate .css file for just this bit
*/
h3 {
  /*text-align: center;*/
  margin-top: 1.25em;
  margin-bottom: 1.25em;
}
```

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = FALSE)

library(lese)
library(ggplot2)

included_lectures <- lectures()

if (!file.exists("slide_check_cache.rds")) {
  stop(paste(
    "slide_check_cache.rds does not exists.",
    "Run `make` and get a coffee, or",
    "Rscript -e 'source(here::here(\"helpers.R\")); check_all_slides()", sep = "\n "
  ))
}

all_slide_check <- readRDS("slide_check_cache.rds")

# Get lectures and reorder them
# Ignores lectures that exist locally but are not listed in included_lectures
lectures <- unique(all_slide_check$lecture)
lectures <- lectures[match(included_lectures, lectures, nomatch = 0)]

# Might as well subset table to only selected lectures
# This allows having a lecture locally for testing but not include them here yet.
all_slide_check <- all_slide_check[all_slide_check$lecture %in% included_lectures, ]
```

This document tabulates the state of the slides in `lecture*/slides/<topic>/` in terms of a) compilability and b) comparison with presumably "known good" versions of the slides located in `lecture_*/slides-pdf/`.

# Latest Commits

Displaying the latest commits of the lecture service repository and each of the currently included lectures (for debugging and race condition mitigation (kind of)).

```{r slide-checker-repo-status}
this_repo_status() |>
  dplyr::mutate(
    branch = sprintf("<a href='https://github.com/slds-lmu/lecture_service/tree/%s'>%s</a>", branch, branch)
  ) |>
  dplyr::select(branch, last_commit_time, last_commit_by, last_commit_summary) |>
  kableExtra::kbl(
    col.names = c("Branch", "Last Commit", "By", "Message"),
    escape = FALSE, caption = "Status of the lecture_service repo (time in UTC)"
  ) |>
  kableExtra::kable_styling()
```


```{r lecture-repo-status}
lecture_status_local(lectures) |>
  dplyr::arrange(dplyr::desc(last_commit_time)) |>
  dplyr::mutate(
    lecture_link = sprintf("<a href='https://%s.github.io/%s'>%s</a>", org, lecture, lecture),
    branch = sprintf("<a href='https://github.com/%s/%s/tree/%s'>%s</a>", org, lecture, branch, branch)
  ) |>
  dplyr::select(lecture_link, branch, last_commit_time, last_commit_by, last_commit_summary) |>
  kableExtra::kbl(
    col.names = c("Lecture", "Branch", "Last Commit", "By", "Message"),
    escape = FALSE, caption = "Latest commits per lecture (times in UTC)"
  ) |>
  kableExtra::kable_styling() |>
  # Increase width of message column as it contains the most text (potentially)
  kableExtra::column_spec(5, width = "40%")
```

If you're missing a lecture here, you might have to add them to or ensure they are downloaded correctly.

# Slide status

```{r slide-status-plot, eval = requireNamespace("ggplot2")}
#| fig.width: 10
#| fig.height: 6
#| out.width: "100%"

check_counts <- do.call(rbind, lapply(c("compile_check", "compare_check"), \(x) {
  all_slide_check |>
    dplyr::filter(lecture %in% .env$included_lectures) |>
    dplyr::count(lecture, .data[[x]]) |>
    dplyr::group_by(lecture) |>
    dplyr::mutate(prop = n / sum(n), check = x) |>
    dplyr::rename(result = dplyr::all_of(x)) |>
    dplyr::mutate(
      result = dplyr::case_when(
        result ~ "Passed",
        !result ~ "Failed",
        .default = "N/A"
      )
    )
})) |>
  dplyr::mutate(
    lecture = stringr::str_remove(lecture, "^lecture_"),
    lecture = factor(lecture, levels = stringr::str_remove(included_lectures, "^lecture_")),
    check = stringr::str_remove(check, "_check$"),
  )

check_counts |>
  ggplot(aes(x = check, y = prop, fill = result)) +
  facet_wrap(vars(lecture)) +
  coord_flip() +
  geom_col(alpha = 2/3) +
  scale_y_continuous(labels = scales::percent) +
  scale_fill_manual(
    breaks = c("Passed", "Failed", "N/A"),
    values = c(Passed = "#1B9E77", Failed = "#D95F02", "N/A" = "#7570B3")
  ) +
  labs(
  #  title = "Lecture Check Overview",
    y = "Proportion of Slides", x = NULL, fill = NULL
  ) +
  theme_minimal(base_size = 16) +
  theme(
    legend.position = "bottom",
    plot.title.position = "plot",
    panel.grid.minor = element_blank(),
    panel.grid.major.y = element_blank(),
    axis.text.y = element_text(angle = 90, hjust = .5)
  )
```


<details>
<summary>Click to expand explanation of table columns</summary>

```
lecture_advml/slides/gaussian-processes/slides-gp-bayes-lm.tex
\___________/        \________________/ \________________/
   lecture                 topic            <slide-name>

lecture_advml/slides-pdf/slides-gp-bayes-lm.pdf
                        \___________________/
                       pdf_static / "Reference PDF"
```

Tables are organized by lecture (first level) and lecture topics corresponding to folders in `slides/<topic>` (second level), with a table per topic.
Topics ignore certain folders like `attic` and `all`.

- **Slide**: The `<slide-name>` with a link to the compiled .tex file (if it compiles).
- **Compiles?**: Whether the .tex file compiles without error using `latexmk -pdf`.
  This can depend on the availability of specific LaTeX dependencies, but assuming they are not the issue, this usually indicates syntax issues or things like `\includegraphics` calls using relative paths that work on Overleaf but not locally.
- **Comparison**: Checking the compiled `<slide-name>.tex` against `slides-pdf/<slide-name>.pdf` using `diff-pdf-visually`:
  - &#x274c;: The check failed surely, either due to differing page counts or because there is no reference PDF.
  - &#x2753;: The check indicates differences on at least one slide, but they may be minor and need manual checking.
  - &#x2705;: The check seems to be fine, the PDF files match.
- **Note**: A comment on the previous column listing the specific reason for a failing check.
  Also enumerates the slide pages where differences are found, and links to a PDF containing a diff of the two files generated with `diff-pdf` (at a lower resolution than the originals).

</details>

```{r data-prep, include=FALSE}
check_table <- all_slide_check |>
  # Transforms the logical columns, i.e. pdf_exists, pdf_static_exists, compile_check, ...
  dplyr::mutate(dplyr::across(dplyr::where(is.logical), \(x) {
    dplyr::case_when(
      x ~ "\u2705",  # Green check symbol
      !x ~ "\u274c", # Red X symbol
      is.na(x) ~ ""
    )
  })) |>
  dplyr::mutate(
    # Insert red question mark emoji instead of red X for easier visual parsing maybe
    compare_check = ifelse(grepl("Dissimilar pages", compare_check_note), "\u2753", compare_check),
    compare_check = ifelse(grepl("No reference", compare_check_note), "", compare_check)
  ) |>
  # Make links out of a few columns
  dplyr::mutate(
    comparison_path = here::here("comparison", fs::path_file(pdf)),
    check_note = ifelse(
      compare_check %in% c("\u274c", "\u2753"),
      sprintf("<a href='%s'>%s</a>", fs::path_rel(comparison_path), compare_check_note),
      compare_check_note
    ),
    compile_note = stringr::str_replace_all(compile_note, "`", "'") |>
      stringr::str_replace_all(r"(\\)", r"(\\\\)"),
    check_note = ifelse(compile_check == "\u274c", compile_note, check_note),
    slide_name = ifelse(
      fs::file_exists(pdf),
      sprintf("<a href='%s'>%s</a>", fs::path_rel(pdf), slide_name),
      slide_name
    )
  )
```


```{r  slide-status-table, results='asis'}
# This is a markdown "asis" chunk that can output raw markdown or HTML and is treated as such by knitr
# Useful to manually assemble tables with sub-headings

for (current_lecture in lectures) {

  tmp <-  check_table |>
    dplyr::filter(lecture == current_lecture)

  cat(sprintf("\n## %s \n\n", current_lecture))

  cat(sprintf("%i topics with a total of %i slides\n", length(unique(tmp$topic)), nrow(tmp)))

  for (current_topic in sort(unique(tmp$topic))) {
    tmp <-  check_table |>
      dplyr::filter(lecture == current_lecture, topic == current_topic)

    cat(sprintf("\n### %s\n\n", current_topic))

    tmp |>
      dplyr::select(
        slide_name,
        compile_check, compare_check, check_note) |>
      kableExtra::kbl(
        format = "html", escape = FALSE,
        align = c("l", "c", "c", "l"),
        col.names = c("Slide", "Compiles?", "Comparison", "Note")
      ) |>
      kableExtra::kable_styling(
        full_width = TRUE,
        fixed_thead = TRUE,
        bootstrap_options = c("hover", "condensed", "responsive")
      ) |>
      kableExtra::column_spec(1, width = "40%") |>
      kableExtra::column_spec(2, width = "10%") |>
      kableExtra::column_spec(3, width = "10%") |>
      kableExtra::column_spec(4, width = "40%") |>
      cat()
  }
}
```