---
title: "Data Snapshot Report"
format:
  html:
    code-fold: true
    code-summary: "Show Code"
vignette: >
  %\VignetteIndexEntry{Data Snapshot Report}
  %\VignetteEngine{quarto::html}
  %\VignetteEncoding{UTF-8}
---

```{r}
#| label: setup
#| include: false
#| error: true

data_loaded <- FALSE
snapshot <- NULL
history <- NULL
history_summary <- NULL
universe <- NULL
snapshot_info <- list(generated_at = NULL, source = NULL)

tryCatch({
  suppressPackageStartupMessages({
    library(etfdata)
    if (requireNamespace("dplyr", quietly = TRUE)) library(dplyr)
    if (requireNamespace("readr", quietly = TRUE)) library(readr)
  })

  snapshot_file <- "vignette_data.rds"
  data_path <- system.file("extdata", snapshot_file, package = "etfdata")
  if (data_path == "" || !file.exists(data_path)) {
    data_path <- file.path("inst", "extdata", snapshot_file)
  }

  if (file.exists(data_path)) {
    snapshot <- readRDS(data_path)
    universe <- snapshot$universe
    history <- snapshot$history
    history_summary <- snapshot$history_summary
    snapshot_info$generated_at <- snapshot$generated_at
    snapshot_info$source <- snapshot$source
    data_loaded <- TRUE
  }
}, error = function(e) {
  message("Failed to load snapshot: ", e$message)
})
```

This report summarizes the cached ETF data stored in the package. The targets
pipeline only downloads missing data or missing date ranges.

## Snapshot Overview

```{r}
#| label: snapshot-overview
#| echo: true

if (data_loaded && !is.null(universe)) {
  cat("Snapshot generated at:", format(snapshot_info$generated_at), "\n")
  cat("Source:", snapshot_info$source, "\n\n")
  cat("Tickers:", nrow(universe), "\n")
}

if (data_loaded && !is.null(history) && nrow(history) > 0) {
  cat("Total rows:", nrow(history), "\n")
  cat("Date range:", format(min(history$date)), "to", format(max(history$date)), "\n")
}
```

## Coverage Summary

```{r}
#| label: coverage-summary
#| echo: true

if (data_loaded && !is.null(history_summary) && nrow(history_summary) > 0) {
  ok_count <- sum(history_summary$ok, na.rm = TRUE)
  total <- nrow(history_summary)
  cat("Complete coverage:", ok_count, "/", total, "\n")
  cat("Missing coverage:", total - ok_count, "\n\n")
  print(head(history_summary, 10))
}
```

## Missing Data

```{r}
#| label: missing-data
#| echo: true

if (data_loaded && !is.null(history_summary) && nrow(history_summary) > 0) {
  missing <- history_summary %>%
    dplyr::filter(!.data$ok | .data$rows == 0)
  if (nrow(missing) == 0) {
    print("No missing coverage detected.")
  } else {
    print(head(missing, 20))
  }
}
```

## Recently Added Rows

```{r}
#| label: rows-added
#| echo: true

if (data_loaded && !is.null(history_summary) && nrow(history_summary) > 0) {
  recent <- history_summary %>%
    dplyr::filter(.data$rows_added > 0) %>%
    dplyr::arrange(dplyr::desc(.data$rows_added))
  if (nrow(recent) == 0) {
    print("No new rows were added in this snapshot.")
  } else {
    print(head(recent, 20))
  }
}
```

## Session Info

```{r session-info}
sessionInfo()
```